An eDiscovery process can touch many areas within your company, and meeting the obligations of the Federal Rules of Civil Procedure (FRCP) regarding eDiscovery processes can be a daunting task. According to the National Law Journal over 90 percent of business documents today are created and stored electronically. When you factor in the many avenues in which corporate data is created and stored, whether it is e-mail, spreadsheets, or electronic documents, the need for adopting a proactive eDiscovery process becomes apparent.
It is well established that not having a formal process towards eDiscovery is a costly and risky strategy. Studies performed by DCIG have shown that costs can approach $2000 per gigabyte of information for collecting and preserving electronic data. This does not include the costs associated with a legal review that a proactive approach to eDiscovery can help mitigate
Traditional reactive approaches to eDiscovery expose companies to skyrocketing costs as they try to wade through the processing and review of e-mail and documents which too often cause delays in determining which documents are subject to discovery. A proactive approach towards eDiscovery shortens the time it takes to identify, collect, process, and analyze electronically stored e-mail and documents thereby reducing costs and delays in responding to eDiscovery requests.
In developing a proactive approach to eDiscovery, there are several areas which a company needs to consider:
1. Is a company better served through outsourcing eDiscovery to a third party? Or is bringing up-front eDiscovery in house through the use of an eDiscovery tool such as Kazeon Systems' IS1200 a more prudent option? When making this decision companies need to consider all issues and risks and decide which model best fits their corporate strategy. Usually the decision is one based on cost and control. How can a company have the most control over its data and implement the eDiscovery process for the lowest cost?
2. Does your company have a good understanding of the laws pertaining to eDiscovery? The revised FRCP governs eDiscovery in legal proceedings. Ensuring there isn't any legal counsel confusion regarding eDiscovery is critical during these times since, in order to comply with FRCP, there are steps that companies need to take including:
If you do not have a good handle on your electronically stored information, then any eDiscovery request is going to be a costly and time consuming effort. Companies no longer have open-ended time frames in which to answer an eDiscovery request and there have been high profile examples in which civil penalties have been levied against companies for delays in responding to eDiscovery requests. One such example as reported by the New York Times was Morgan Stanley's $15 million agreed upon fine with the SEC for failing to produce e-mails in a timely manner.3. Developing an adequate records management program.
4. Finally, companies should develop policies and procedures to ensure they have a plan in place when eDiscovery occurs. This should be a combined effort between corporate management, corporate IT and legal counsel to ensure a good understanding of the challenges in responding to eDiscovery requests. A consistent approach to these types of requests will go a long way in making sure management, IT and legal are all on the same page when a company needs to perform an eDiscovery.
Legal fire drills generate a lot of activity and sometimes solve the problem but they are no longer a viable substitute for a well managed and documented corporate eDiscovery process. Companies must understand that a reactive approach to eDiscovery results in unacceptable delays and could lead to critical omissions and/or revelations. These may damage the organization's reputation in the market while also increasing costs and taking up valuable time. By following the above suggestions your company can begin to implement a thoughtful and prudent pro-active eDiscovery infrastructure that will benefit not only your ongoing information management initiatives, but help you respond to reactive eDiscovery requests with unprecedented completeness and accuracy across your enterprise.
]]>
Though Quantum announced the DXi7500 some time ago, today marks general availability of the DXi7500 that it views as meeting today's enterprise concerns. There are a number of features that Quantum includes within the DXi7500 to scale into the enterprise including:
One of the more innovative features of the DXi7500 is its adaptive approach to deduplication. It appears that when Quantum was looking at how to best approach deduplication for the enterprise, it did not to take sides in the "inline" versus "post- processing" deduplication debate and instead gave the customer the flexibility to choose which approach best suits their needs.
This is particularly valuable to companies that have a mix of backup jobs with both low and high performance characteristics and that need more than one alternative when deduplicating backup data. Other products only offer one choice for deduplicating data - "inline" or "post-processing" - but not both. Using Quantum's adaptive deduplication approach, however, companies can match the form of deduplication to the characteristics of their backup job or even the nature of the data contained in the backup jobs. Configured this way, the DXi7500 can match the deduplication approach to the requirements of specific backup jobs.
Matching the deduplication approach to the type of backup job, or even data within the backup job, delivers faster ingest speeds during the backup window. However, because the DXi7500 also gives companies the option to create partitions and assign specific deduplication methods to these partitions, companies can designate which partitions use what specific type of deduplication. In those circumstances where companies expect moving backup data to tape immediately or backups contain a high percentage of new data, companies can take advantage of its "fully deferred" deduplication policy that postpones the deduplication until the backup is fully complete.
Quantum also looks to give companies a compelling reason to use the DXi7500 as the target for receiving replicating data from DXi3500s and DXi5500s as well as using the DXi7500 as a foundation in building a company's disaster recovery (DR) strategy. By deploying DXi3500s or DXi5500s at remote offices, companies can replicate data from these appliances back to a central DXi7500 in their primary data center. Once the data is replicated and centrally stored on the DXi7500, companies can then optionally place a DXi7500 at their disaster recovery site and then replicate data from the DXi7500 in their home office to this secondary site.
The DXi7500 also provides a mechanism to copy data to removable media (tape) for those companies that do not plan to replicate data to a remote site or have long term archiving or compliance requirements. The DXi7500 can either use its own software to copy data from disk on tape or companies can optionally use Symantec's NetBackup 6.5 that recognizes the DXi7500 and can manage the migration of data from disk to tape.
Of course, a final prerequisite when positioning a storage system like the DXi7500 at the enterprise core is to address enterprise concerns about high availability and reliability. The DXi7500 provides dual RAID controllers, dual redundant power and cooling, and hot replaceable components. Dual-node DXi7500 systems eliminate all single points of failure by using dual DXi controllers, active-active failover for all hardware, and cluster-aware software components that fail-over as needed.
Quantum is one of the first deduplication vendors to make the jump from the midrange market with its DXi3500 and DXi5500 backup appliances into the enterprise space with its DXi7500. On the surface, Quantum appears to have put all of the features into the DXi7500 in order to succeed: a highly available and reliable system, careful navigation of the deduplication debate through adaptive and fully deferred duplication and the use of the same replication software across its midrange and enterprise appliances.
This should work in Quantum's favor near and long term. Despite the fact that the Quantum is somewhat late to market with an enterprise level deduplication system, deduplication is still just getting started, not tailing off. By introducing one of the first core-to-edge global deduplication and replication schemes in the market, Quantum should find ample opportunities for companies that are eagerly looking for a single vendor to meet all of their disk-based data protection and disaster recovery needs.
]]>Yesterday I completed my quick road trip to Chicago to attend TechTarget's annual spring Storage Decisions conference returning home last night. Here are some the highlights from my day 2.
I started out the day with an hour-long briefing with Xiotech's CTO Stephen J Sicola and Storage Architect Peter Selin. Xiotech has been talking up a storm about the ground-shaking importance of its new Intelligent Storage Elements (ISE) ever since Xiotech announced it at Storage Networking World about a month ago. However Xiotech and I have not had a chance to connect for me to take a close look at its architecture so Stephen and Peter spent some time talking me through it.
One of the factoids I found most intriguing was the history (at least as Steve tells it) why Xiotech (and Seagate behind the scenes) felt obligated to go back to the basics in designing the ISE that underlies its new Emprise stroage system. One of the more interesting aspects to the story was the history of placing disk drives into storage systems. Apparently when disk drives were first placed into storage systems, they were not designed them for vertical insertion - always horizontal. So when disk drives were placed vertically in storage systems to optimize rack space, they started failing more frequently.
Another key problem had to do with mounting and cooling the disk drives. Again, disk drives were designed for mounting in stable (non-vibrating) racks as standalone units with ample air flow for cooling. However, when putting tens or hundreds of disk drives into a rack, not only is air flow around the disk drives reduced, but the vibration of all of these spinning disk drives in the same rack is amplified leading to higher disk drive failure rates. So the disk storage systems have compensated over the years by making tweaks in their firmware and controllers to offset these variances and minimize the impact of failures.
Xiotech's Sicola felt it was time to go back to the drawing board and re-examine the design of everything from the disk drive firmware to how they were mounted in storage systems to the controllers managing them. He started this process nearly 6 years and the result is the ISE found in Xiotech's Emprise storage systems. Key changes it makes are more stable mounts for disk drive placement and replacing the disk drive's native firmware with its own firmware for more pro-active monitoring and the transmission of storage system reports to Xiotech.
Though there were many others, sending the activity reports to Xiotech caught my attention because Xiotech will now monitor activity on your systems and notify companies not just when drives fail, but warn them when it detects abnormal activity on their Emprise system that may contribute to degraded application performance. For instance, if a company places a high performance Oracle database on SATA disk drives, the reports sent back to Xiotech should detect this activity and Xiotech should in turn warn the company that not only should its Oracle database not reside on SATA disk drives, but that this level of activity could lead to degraded performance and SATA disk drives on the system failing prematurely.
So what do all these new features mean for users short and long term? Because Xiotech makes the Emprise more resilient, they have extended the warranties on their systems from 3 to 5 years while its upfront costs are comparable to other systems. This should allow companies to depreciate these systems out over five years rather than three. This can lower quarterly depreciation costs and, since the underlying disk drives are theoritically more reliable, there is a lower chance of disk drives failing and hence less risk to your applications.
The main question companies need to ask themselves about Emprise is not about its stability and reliability but did Xiotech over-engineer this system? Five years is a long time in the technology industry and can span as much as three generations of technology improvements (assuming new technology is introduced every 18 months). This can leave a company with book value on a 3 year old storage system should a need to upgrade it to more current technology. This could require the company taking a financial hit on the books even though the Emprise is still a viable storage system. Overall, though, Xiotech's Emprise should give companies pause about their current vendor's storage system and think more deeply about how their current storage systems are archtitected and if going from a 3 to a 5 year warrantly makes sense.
My next meeting was with Omneon's Director of Storage Marketing, Dave Frederick. Omneon is a 10-year old, $120 million storage company primarily dedicated to providing storage for the broadcasting industry so I inquired of Dave why his company's sudden interest in attending Storage Decisions. He said that more Fortune 500 and Fortune 1000 companies are now broadcasting video internally and this is creating a new demand for storage systems specifically designed for the broadcasting industry.
So I queried Dave further to understand further how high transaction environments differ from broadcasting since both call for near 100% availability. Dave explained that there are two fundamental differences between broadcasting and high transaction environments. Broadcasting accesses data sequentially while high transaction environments tend to access data randomly. However, the larger difference is that if there are pauses in high transaction environments (even milliseconds), the transaction can be resent. This is not so in broadcasting. It even one frame is missed (30 frames are sent every second), you don't get a second chance and those types of misses (called black spaces) result in missed SLAs and loss of revenue for broadcasting companies.
It is in this way that Omneon's MediaDeck Integrated Media Server storage differentiates itself from competitive products. Though it uses a grid storage architecture, it also includes an out of band component that verifies each frame as it is encoded and decoded so that when a broadcast is sent out, it streams the video without a black spaces.
Finally, my other notable meeting for the day was lunch with representatives from the LTO consortium: Quantum's Product Marketing Manager, Tom Hammond; IBM's Senior Program Manager, Bruce Master; and HP's Product Marketing Manager, Rick Sellers. Most of our conversation focused around how the use of tape is changing in environments and that while disk is becoming the primary target for backup, companies still need to exercise some caution about using disk exclusively for backup. All of us were aware of recent examples where companies had both their primary, secondary and, in one case, even a tertiary DR site affected by disasters that required the use of portable media in order to recover their environment at still another site.
]]>Finding key terms and phrases buried inside of mountains of recorded phone conversations, voice mails and IM chats can devour discovery budgets and send counsel crying 'undue burden' to the court. There seem to be two dominant speech analytics methods: phonetic indexing (first brought to eDiscovery by Nexidia) and transcription or speech-to-text (long dominated by Autonomy's engine which supports both methods). Phonetic search renders sound wave forms into simplified strings of phonemes that can be indexed and searched. This makes the technology effectively content agnostic, but makes it challenging to integrate with text based search. Speech-to-text has been the foundation for automated conceptual search and improvements in speaker recognition technologies have also increased the value in what was effectively raw dialogue.
Knowing that one can search digitized conversations, the next question is can users effectively search everything within the enterprise system from unified federated search? There is little doubt that the archiving systems are aggressively pursuing acquisitions, partnerships and development to enable ingestion and indexing of every conceivable data stream. All of them started with email back in the late 1990's. For example, Symantec doesn't have audio, but jumped ahead with early products to handle IM, file shares, Sharepoint through merger and acquisition.
Mergers inject complexity by requiring integration of technology, services and cultures. Having experienced the Symantec-Veritas integration first hand, it always surprises me to see pundits and bloggers jumping all over the expected personnel departures. Moreover, other restructurings that occur in the wake of big M&A moves. Instead of looking at who departed, which products are End-of-Lifed and which partners jump ship, I think that we should look at who stays, fundamental IP integrations and new solution offerings to get a better idea of where the new joint entity is headed.
As a former customer of Zantaz's Introspect, I watched press releases closely after the Autonomy acquisition last July. I was surprised to see how quickly all of Zantaz's products were integrated to the IDOL search architecture. A two to three month integration tells me that the business units got the resources and investment needed to make necessary changes under the hood. The back channel continues to be positive. For example, after speaking with two big Introspect customers at a conference last week, it was clear that Autonomy had greatly improved search speed and performance. The back end administration and database complexity still seems to be an issue with some customers, but systems that offer enterprise scale will require enterprise level investments in architecture, support and consistent project management to succeed. Perhaps that is at the root of the shift within the Zantaz channel to go up market and target larger sales?
If we are going to demand that large public corporations make all of their communication data streams accessible and manageable to respond to the new FRCP requirements, then we have to expect further consolidation in the maze of different applications used today. Discovery, ILM, Retention Management, Enterprise Content Management and all the other flavors of alphabet soup are just ways of saying that companies are responsible for administrating their information assets, including formerly transient forms such as audio, IM, PIN-2-PIN, etc.
So what is a smaller player to do? Well there are quite a few new SaaS offerings to consider. The model makes sense for many smaller companies with 1-100 users, especially if they are a service company that wants to minimize overhead and infrastructure. If you have good access to bandwidth or your employees are geographically diverse, then chucking the entire messaging platform can be very attractive.
But outsourcing your informational assets may not be an option for regulated verticals or public agencies with FOIA requirements like the Florida Sunshine Laws. A
So the under-represented segment of the market needs an archiving and discovery solution with a low entry and implementation threshold. Moreover, it has to be easy to use and administrate as smaller entities may not have dedicated personnel with deep skills. Some of the archiving players seem to have focused on meeting these challenges. Estorian's Looking Glass Interactive Archive is sold on a subscription model with an entry point as low as $400 per month. The system leverages the Exchange's 'Receive As' functionality to intercept active traffic without using Journaling. This also enables them to scan the mailbox and attached PSTs via MAPI call to dynamically track the folder location and other user actions on the items in the mailbox. This makes it easier to roll outfor smaller companies, because journaling and stubbing are de-emphasized.
This week I was back on the road again for a short trip to Chicago (short for me anyway since I am from Omaha) to attend the annual spring Storage Decisions conference put on by TechTarget.
Despite some of the rumors that were floating around that user attendance was down and not as many vendors were in attendance, I did not necessarily find that the case. I ran into Lindsay Mullen, TechTarget's event coordinator, in the exhibit hall. She said that about 550 end-users had registered for the event and that she still expected another small influx of end-users on Wednesday. Also, in talking to other analysts and TechTarget editors who were in attendance, they said that all of the break-out sessions were full. The number of vendors exhibiting did, however, seem to be down from years past but not as much as I anticipated. Off-hand, I am guessing that the number of vendors exhibiting was about two-thirds of past events.
For me, the first day had a decided NEC flair to it. I arrived mid-afternoon on Tuesday and met first with NEC's Karen Dutch. Though I know Karen quite well and she and I speak often, there were a couple of salient points that came out of our briefing. The first was that as she speaks to end-users, she finds that there is a tendency among end-users to lump all disk-based backup products into the virtual tape library (VTL) category. Having been guilty of making that assumption myself in the past, it's easy for me to believe that other end-users are falling into the same trap. However, based upon my observations and trends I am seeing the market, I would venture to say that there is a definite trend away from VTLs and towards using disk-based appliances configured as NAS. Even vendors like Overland Storage who predominantly provide VTLs are examining the possibility of releasing a NAS-based interface.
The other topic that Karen and I discussed at length was the willingness of enterprise users to accept the NEC HYDRAstor's disk-based NAS interface. NEC definitely has the enterprise in mind with the HYDRAstor but I have wondered if enterprises were ready to make the jump from VTL to a disk-based backup appliance configured as NAS. Karen said she had had that concern as well but so far that concern has not born out. Though NEC has a version of the HYDRAstor that is configured as a block-based VTL ready for testing, so far it has not found the need to bring a VTL-based version of the HYDRAstor to market.
Following my briefing with Karen, I stopped by the exhibit hall for a couple of hours to see who was exhibiting and catch up on the latest industry buzz. While grabbing a bite to eat, I ended up sitting with Greg Schulz from Storage I/O, We were sitting across from the HP booth so it naturally followed that we would talk about HP's acquisition of EDS. Having heard about the acquisition on Monday night, I already had a day to mull it over and thought the acquisition made sense. Greg concurred. He felt that HP had a solid insourcing model for those companies who wanted to keep most of their computer management in-house but still needed the more traditional break/fix support that HP current offers. By acquiring EDS, HP now gains access to an outsourcing model for those companies that want to outsource components of their IT infrastructure that are not strategic to their core business. These companies can now turn to HP for both of those functions, something they could not do in the past.
Next up was the evening event that was hosted by NEC (hence the NEC flair to my Day 1 on Storage Decisions). During this event, I was seated next to an end-user from Davenport, IA, and we got to talking about what I saw as the hottest thing in storage right now. I told him that the evolution of data protection software to data management software is probably the most encouraging trend going on. Whenever I speak to users of Asigra Televaulting or the CommVault Simpana Suite, these users are almost universally telling me that they have moved beyond the day-to-day fire drill of troubleshooting backups to focusing on data management and data recovery. That statement obviously struck a chord with this user. He uses Tivoli Storage Manager (TSM) and he said that he wished he was at that point with backup because he still spends a lot of time struggling with backup and recoveries.
Stay tuned for more insight and info on Storage Decisons Chicago in tomorrow's blog entry.
]]>Virtualization is having the same type of revolutionary impact on corporate data centers that using disk as part of the data protection process has had. Though companies utilize virtualization in many ways, server virtualization is where its impact is most apparent. VMware, Virtual Iron and Xen are contributing to server virtualization's current growth while Hyper-V in Microsoft's Windows Server 2008 is poised to join the virtualization fray this summer.
In the face of these fundamental shifts among corporate data centers in server data protection and virtualization, data protection software needs to do more than just adapt. It needs to embrace backup-to-disk and server virtualization in order to transform data protection software into an information recovery platform. That is exactly what today's 8.0 release of Asigra Televaulting brings to the table in the following three ways:
First, Televaulting 8.0 introduces on-demand grid computing. Televaulting has used an agentless architecture since its inception. However its DS-Clients (its agentless backup servers) were configured as one-to-many. This architecture precluded a single DS-Client from helping other DS-Clients complete backup jobs in enterprise shops with large numbers of servers - physical or virtual.
Using Televaulting 8.0's new on-demand grid computing architecture, a parent DS-Client is now created that is aware of the other child DS-Clients residing on virtual or physical machines in the backup environment. As child DS-Clients complete backup jobs assigned to them, they notify the parent DS-Client that its jobs are completed and that it can help other child DS-Clients complete their assigned backup jobs. The parent DS-Client then re-distributes jobs from busy child DS-Clients to the idling child DS-Client so it can help in the backup process.
Second, as I go to storage conferences and speak to end-users, the wide-scale acceptance and adoption of wide area replication astounds me. The size of the company no longer matters. Whether they as small, medium or large, companies want this feature because they understand the age of recovering data anywhere, anytime, and anyplace is upon them.
Multi-directional Data Center Replication in Televaulting 8.0 addresses this growing enterprise need. Asigra Televaulting has for years deduplicated and then replicated data from remote sites back to its central site. However, enterprise companies may have multiple data centers with a need to replicate data between these sites. The Multi-directional Data Center Replication in Televaulting 8.0 capitalizes on the new on-demand grid computing architecture to create an "N+1" configuration. Using "N+1", DS-Systems (the parent Televaulting server into which child DS-Clients feed their data) are aware of each other, sending and receiving backup data to one another. A DS-System can also take over for a DS-System at another site should it fail or need to go off-line for maintenance.
Finally, Asigra Televaulting has always used disk as its primary target for backup but treated all files as the same and worked on the assumption that there was only one tier of disk. Televaulting 8.0 makes no such assumption. Instead, it leverage its knowledge about the files it is protecting (age, size, date last accessed, etc.) plus it now understands that multiple tiers of disk (Fibre channel, SATA, MAID) may exist.
By using this knowledge about the data and disk storage systems, Televaulting 8.0 introduces intelligent archiving as a feature set by placing a server's most active data on FC disk to improve performance during backups and recoveries. However it can store infrequently accessed data on SATA or MAID disk storage systems to take advantage of their lower disk costs and higher capacities while minimizing power consumption.
Asigra recognizes that a fundamental shift is occurring in data protection and it is about more than just supporting or adapting to backup to disk or virtual machines - it is about embracing them. Televaulting 8.0 takes this next logical step in its product evolution by not just adapting to these technologies but making them core to its next generation of prodcut. The introduction of on-demand grid computing, multi-directional replication and intelligent archiving in the newest version of Televaulting 8.0 demonstrate Asigra is committed to helping its clients usher in a new era of information recovery.
]]>Some of the problems are obvious. Retaining more data drives up storage costs across the board from extra capacity needed on production storage systems to the need for more capacity in backup. Though deduplication helps to take some of the sting out of disk-based storage costs and using disk as a backup target helps shorten the backup windows, this is only a short term reprieve in coping with existing corporate data and storage management problems, not the final solution.
Most companies fail to understand the financial, technical and legal liabilities that keeping this amount of data online presents to their companies. Andrew Reichman, a Senior Analyst with Forrester Research, conducted studies over the last couple of years that illustrates some of the risks that unmanaged data presents to companies. His findings with Forrester Research included:
Actual experiences and expenditures will vary by company but these findings clearly identify/illustrate that companies are approaching a point where they must pro-actively manage their data and start to separate production data from stale or infrequently accessed data. Moving this data from primary storage to secondary reduces backup windows since less data is backed up during full backups. Companies can also reduce the amount of money they pay for secondary storage since, instead of procuring high cost Fibre Channel storage systems, they can purchase lower cost, higher capacity storage systems to house this type of data.
However before anyone runs down to CompUSA to start buying hard drives and servers, companies need to put some thought into the type of storage system that they are going to use to host archived data. Companies should also weigh the wisdom of using a product from their existing storage system vendor's portfolio of storage systems as their storage system for archived data. The challenge that most companies will find in this situation is that while most storage systems using off-the-shelf SATA disk drives will cost less than FC disk drives, companies cannot forget to factor in the technical challenges, legal liabilities and financial costs associated with long term data archiving and retention.
Selecting a system that has a scalable and flexible architecture and also satisfies external legal compliance issues by retaining data for the appropriate time without keeping it too long; thereby exposing companies to new risks is a separate issue. In the next blog entry, I'll take a closer look at:
One thing that struck me was that Compellent (NYSE: CML) users really understand what a game-changing technology that virtualization is. I sat through 2 or 3 presentations during the two days of the conference (May 7 - 8) and also met with a fair number of users (~10) between sessions, over meals and at the evening events and all of them were pretty stoked about the capabilities that virtualization in general and Compellent specifically delivers.
Compellent's Data Progression (automated tiered storage) was the virtualization feature that its users spoke most highly about. One user I spoke with over drinks who was from
Compellent's Dynamic Capacity (thin provisioning) feature was given a lot of attention at the user conference but none of the users I spoke to seemed to be using it - or at least it never came up in conversations that I had with them. It might just be that they assumed I knew they were using it since the Dynamic Capacity feature is part of Compellent's
Replication was clearly on the mind of almost every user whether they were presenting at the show or merely talking with me privately. It seems a fair number of its users are taking advantage of Compellent's Storage Center replication functions, Remote Instant Replay (asynchronous replication) and Data Instant Replay (snapshots), in some way even though these software features are add-on licenses to the core software. This trend confirms my suspicions that fast recoveries are becoming more important for companies and the end-users they support.
However not all of the news around replication was positive. Most users had no problems using Compellent to replicate data locally or remotely but when it came to providing consistent recoverable snapshots in conjunction with applications, the news was somewhat mixed. During one user panel, Bill Moss, IT Director for Moss Construction Managers in
Also at the conference, I had the opportunity to meet with Bruce Kornfeld, Compellent's VP of Marketing, and Larry Aszmann, Compellent's CTO. The main item Bruce and I discussed was how Compellent licenses its software. What distinguishes Compellent from most of its competitors is that it licenses its software by spindle (per disk drive). Its core licensing includes Dynamic Capacity (thin provisioning), LUN security, boot from SAN, some base level reporting features and email home support. This licensing is based on a single controller with one shelf of 16 disk drives. As companies grow their Compellent system, Compellent sells disk drives and licensing in what it terms as "8-packs". Additional software features that users can optionally license with larger systems include its Data Progression, Data Instant Replay, Remote Instant Replay and Fast Track features.
In the brief meeting I had with Compellent's CTO Larry Aszmann just before I exited the conference, I gleaned two pieces of new information regarding Compellent's manufacturing process and its commitment to its VARs. In regards to manufacturing, Compellent primarily uses off-the-shelf components in the construction of its systems. This removes from Compellent many of the traditional manufacturing concerns that other storage system providers need to manage.
Aszmann also said that Compellent sells 100% of its products through the channel and has no plans to go direct. He has seen other storage systems vendors do that which has ultimately undermined the relationship with their VARs. Because Compellent does not sell direct, VARs are much more transparent with Compellent about their business dealings since they are less worried about Compellent cutting them out of deals later on. Aszmann says this level of transparency is helping it as a public traded company because it can remain very accurate (about 90% on target) with each quarter's sales forecasts.
]]>I first met Mr. Paulk at the Fall 2007 Storage Decisions conference in New York City and interviewed him shortly thereafter for a blog entry that appeared back in October 2007. However six months have passed since that interview so I followed up with Mr. Paulk to get an update on how his installation of the NEC HYDRAstor was performing, since Paulk was still using a beta version of the NEC HYDRAstor software when we last spoke.
Paulk revealed that he is now in full production with the production code loaded on the NEC HYDRAstor. However he is still using the same hardware configuration (two Accelerator Nodes and four Storage Nodes) that he started out using due to the high deduplication ratio that he is achieving with the HYDRAstor.
Last fall he was achieving a 17:1 deduplication ratio and hoped to eventually achieve a 35:1 ratio. Six months later, his deduplication ratio is now approximately 39:1 which has mitigated his need to buy additional capacity and has driven his cost/GB down to approximately 70¢/GB. "It's like getting 390 TB for the price of 10 TBs," says Paulk.
He also has not found it necessary to add more Accelerator Nodes into his HYDRAstor configuration. Though he has nearly doubled the number of servers he is backing up on a nightly basis (from 13 to 21 servers), he is achieving about 3 Gbps of throughput across his two Accelerator Nodes.
I then asked him, "What are the biggest benefits that you have experienced since you started using the HYDRAstor?" There were four benefits he cited:
In the next few months, Paulk plans to archive about one million documents to the HYDRAstor, which will consume about another 1.4 TBs of storage. What he is curious to discover is how this will impact his deduplication ratio. These one million documents are currently paper documents that need to be scanned so he is anxious to find out if he will achieve the same high deduplication ratio when archiving the paper documents that he is experiencing with his backup data. My plan is to catch up and speak with Mr. Paulk again this fall so DCIG can share some more of his story and experiences at that time.
]]>Today and tomorrow I am putting on both my reporter and analyst hats. Living in Omaha, NE, I am only a hop, skip and jump away from Minneapolis, MN, so I took the opportunity to drive up here to attend Compellent's annual C-Drive user conference that runs from May 6 - May 8 and do some live, on-site blogging about my experiences while I am here.
Already a few notable items to report from last night's customer reception and this morning's opening presentation.
At the customer reception at Brit's Pub in downtown Minneapolis, I ran into Scott Horst, Compellent's Director of Marketing, and had a chance to chat with him for a bit. He said that about 100 users were registered for the event which represents nearly 10% of their current customer base since Compellent forecasts hitting the 1000 customer mark yet this year.
Also at the customer reception I had the opportunity to meet Sonia St. Charles, the CEO of the Davenport Group, a Minneapolis based VAR. What was noteworthy about this meeting was that she was one of the first VARs if not the first VAR I have met that has embraced Web 2.0 technologies and has redesigned Davenport Group's entire website with a focus on social networking.
She says her 19-year-old son was part of the motivation to push her company to adopt a Web 2.0 format. She finds that the next generation of storage administrators (35 and under) are not accustomed to being called on by sales reps or willing to wait. Instead they are spending a few hours searching the Internet and getting up to speed on technologies so they can make more informed buying decisons. She sees this as key to her company's future in helping her educate and inform current and potential Davenport Group clients.
This morning (May 7) Compellent kicked off the day with a presentation by its CEO Phil Soran and "Green" was a major part of the theme. The first "Green" was the color that Wall Street types like (as in greenbacks). Some highlights that he shared from Compellent's past year included:
The more popular notion of "green" also appeared in two of the key trends that Soran sees for storage in the remainder of 2008 which include:
I guess up to this point I have been somewhat skeptical about the whole green initiative and still some skepticism remains. Based on Soran's follow-on comments, companies are going green not because they are having any sudden pangs of guilt about wasting too much power or having a carbon footprint that is too large. Rather, power costs are going through the roof and/or there is a real shortage of power.
Soran mentioned one Compellent customer, the US Capitol Police, that had to virtualize their infrastructrure and go "green" because they couldn't get any more power. They used that as an incentive to virtualize their data center thereby making it more green and now actually have power to spare by more efficiently using their resources.
A question and answer period with the audience then followed and here were some of the responses that Soran and other members of the Compellent executive management team had to the audience's questions.
Q&A
When will TB drives be available? In final qualifications right now and Compellent expects them to be released in the next 6 weeks.
Time line on availability of SAS drives? Been testing SAS for 2 years. The drives are fine but waiting until scalability is there. Currently there are limitations in how many drives can be in a loop. It is a technology Compellent will support but not in the immediate near-term (next 6 months).
Solid State? Some customers have solid state in their boxes and Compellent supports solid state drives on a case-by-case basis. There are different kinds of solid state: some are slower than others; others you can only write to a certain number of times. Soran expects them to be adopted first in enterprises with $100 million applications but sees them as problematic to deploy now since customers are still unwilling to ante up for them.
(Side note - I sat in on a customer panel after Soran spoke and there might be some disconnect between Compellent and its customer base on this message. Both of the users on-stage were their highlighted speakers of the C-Drive conference - their pictures are plastered everywhere including Compellent's web site - and both of these presenters indicated they would pursue SSDs if Compellent made them available.)
Encrypted drives? First initiative is to encrypt data stored to removable drives and later to encrypt other drives if demand arose. This is not a 6 week time frame but near future.
How much of user base has upgraded to 4.0? ~250 customers; performance benefits have been good with those users upgrading from earliers releases to 4.0 reporting substantial performance benefits.
Dedupe? Nothing immediate. Thin provisioning and boot from SAN (one volume for all of your server boots) are technologies that Compellent is using now to address current customer concerns but are trying to figure out most logical fit for this technology. They see archive as the most likely fit for dedupe.
Editor Notes: Edits were made to this blog entry on 5/9/08 at 7:00 am CST to make for better reading and correct some grammatical mistakes.
]]>Microsoft's VSS works by using three pieces a requester, writer and provider. If you have heard about this, then skip to the last paragraph and wait for part 2 due later this week. The requester is your traditional backup solution. Requester's send collection inquiries to the application you want to protect. This application must understand the collection inquiry sent to it by the requester and needs a writer designed to support the application data and data types. The writer is written by the application developers, not the backup vendor, to ensure the most stable and consistent recoveries.
Anyway, what caught my interest on his May 1, 2008, broadcast was that he took on the Technology sector and he said that it is time to sell traditional technology stocks. At that point I began to listen more carefully because I have tended to disagree with his observations about technology in the past. Also, since my entire career is centered on covering technology in general and storage technologies specifically, I was curious about how in touch he was with this sector.
To summarize, Cramer feels that many technology companies in Silicon Valley have forgotten their engineering roots and are no longer providing game changing technologies. They now have become so focused on sales and have so many sales people running around trying to sell their products that their engineering departments have gotten lost in the shuffle. As a result, they are no longer focused on delivering products that solve today's immediate problems. Instead they are busy re-packaging yesterday's technology and re-branding it in such a way that it sounds like it solves today's problems.
Having worked as a storage engineer in a Fortune 500 and being involved in the day to day operations as well as having input into its storage buying decisions, I can certainly see his point of view. Sales representatives from traditional technology companies would constantly call on me selling products that did little or nothing to solve the problems that I was trying to solve at that time.
So in the last week I stopped by the place where I used to work just to see how things were going and little has changed except one thing: the people with whom I used to work. All of the people who bought into these 10+ year old technologies that solved yesterday's problems are pretty much gone. And by gone, I do not mean that they were promoted to better paying jobs or leadership roles within the organization. They were gone - as in laid off. Even more ironic, the current regime is now taking the same approach as the previous generation - trying to make old technologies solve today's problems. In that respect, my feelings echo that of Cramer's - "They know nothing!"
I say this not to disparage my co-workers or even traditional technology companies. But when I look at the type of problems that companies face today and how they are trying to fit square pegs (traditional technology) into round holes (today's problems), I have no doubt in my mind that a tsunami of change is gathering that will sooner or later sweep through enterprise organizations.
So do I agree with Cramer? Partially. He is right in the sense that traditional technology companies are behind the curve and that sales departments have overtaken engineering departments in too many of these companies. But I disagree with him that no innovation is occurring in technology - it's just from his vantage point he still can't see it.
I consider Cramer a good barometer of how close we are to a revolution of technology in general and data protection specifically. Cramer sees technology from a bird's eye view whereas I am up close and personal with it. So when I look at products like Asigra Televaulting and its Information Recovery Management platform, I know that this is one product that disproves Cramer's notion that no innovation is occurring in technology. Some of the cool, innovative features that Asigra is about to bring to market I will get into in a forthcoming blog entry.
]]>Any time one looks at midrange backup appliances, the appliances are almost always NAS based. When configured this way, the backup appliance is attached to the local LAN, it appears as a filer server to the backup server and files are backed up to a folder on that appliance. Though I initially called to speak to Overland Storage's Senior Product Manager, Jeff Graham, about REO's Dynamic Virtual Tape (DVT) technology, I first wanted to get some clarification on why Overland Storage's REO-series appliances are configured as Virtual Tape Libraries (VTLs) rather than as a NAS-based appliance.
Jeff provided a couple of reasons as to why Overland Storage currently configures its REO-series appliances as VTLs:
The first reason has to do with performance. Overland Storage has found that its VTLs consistently outperform NAS-based appliances. On NAS-based appliances you have to introduce a file system on the appliance which consumes some server overhead (processing and memory) that a block-based VTL does not consume. However, the larger benefit of using a REO is that it is optimized for storing the large sequential blocks of reads and writes found in backup traffic. Using a NAS, these blocks will fragment over time as the system distributes these blocks of data across the back end disk drives. Using a REO-series VTL, all of the blocks of data are kept together on a specific virtual tape cartridge that helps prevents this type of fragmenting from occurring.
The other reason is manageability. Using a NAS-based backup appliance, folders are created on the appliance's file system that is then used to store the backup data. The problem that can emerge in this scenario is that as if and when the folder fills up with data, the backup job can fail or hang. To allow the backup job to continue, administrators may need to increase the folder's size, manually direct the backup software to backup the data in another folder or manually give the OK to the backup software to overwrite older data in that folder.
Using a REO-series VTL, administrators have a couple of options within the backup software that minimize the likelihood of backup job failures. They can configure the backup software to migrate data from virtual tapes to physical tapes so as the REO VTL fills up, the backup software recognizes this and automatically moves the data from virtual tape to physical tape and then deletes the data on the virtual tape. Alternatively, companies can add more disk capacity to the REO and create more virtual tape cartridges for additional storage capacity.
Bottom-line, using a REO VTL companies do not need to change their backup structure. The backup software recognizes and treats the REO VTL like a physical tape library which minimizes the number of changes that companies need to make their backup infrastructure.
Now, having said all of this, Graham says that Overland Storage is investigating the possibility of adding a NAS option to their backup appliance. The rationale behind this move is that despite the fact VTLs are better suited for backup, there are still market forces at play. A NAS-based backup appliance is simpler for companies to understand since it does not require Overland Storage or its resellers to explain what a VTL is or how it. However, Graham did not indicate if Overland Storage actually plans to move in this direction or, if they are, when such an appliance might be released.
In a future blog entry, I'll delve more deeply into Overland Storage REO's DVT technology, how it works and what makes it unique in the market place.
]]>Bringing backup data from remote and branch offices back to a home office is a particularly thorny problem that enterprises continue to face. Directly sending nightly full, incremental or differential backup jobs over a wide area network (WAN) connection back to the home office can saturate the WAN link and cause backups to exceed backup windows and result in failed backups. However the current procedure of backing up data to disk or tape at the remote site perpetuates the problem of how to most efficiently and securely transmit backup data back to the home office or disaster recovery site.
Placing disk-based backup appliances with data deduplication technology at the remote or branch office can be a good first step towards resolving this problem. Bt placing these appliances at the remote site companies keep backup traffic on their local LAN and off of the corporate WAN and backups complete more quickly and within the backup window. Further, the deduplication feature reduces the amount of backup data stored on the appliance so it can keep more data for longer periods of time.
Still, adding disk-based backup and deduplication to remote sites doesn't solve the challenge of consolidating business data within the core data center (home office) or ensuring all corporate data is secure for disaster recovery/business continuance purposes. In this case, one needs to ensure that the offering provides software that replicates data from the remote office back to the central office.
For companies with only one remote office or that just need to replicate data to a disaster recovery site, a number of appliances provide replication to like appliances at the remote site. Since the data is deduplicated before it is replicated, the amount of data that needs to be replicated from the remote office to the home office is minimized and only net new data needs to be replicated back to the target site. This also lessens the network bandwidth requirements and administrators can configure the replication to occur during periods of low network activity.
The difficulty that arises is when companies have multiple or global remote offices and each has a deduplicating backup appliance. If the deduplicating backup appliance only supports a one-to-one replication configuration, the company may need to purchase enough backup deduplicating appliances for the home office to match the number of deduplicating appliances that they have in the remote offices. This approach is very costly, consumes additional data center resources (power, cooling and floor space) and quickly becomes management intensive, especially in remote offices without adequate IT support.
Companies in these situations need to identify vendors that offer a data deduplication architecture that supports a many-to-one replication model and that can also globally deduplicate the data once it receives the data from the remote sites. By supporting a many-to-one configuration, companies only need one deduplicating backup appliance to receive the data from all of its remote sites. The global deduplication feature is needed since it can further reduce the amount of data that companies need to transmit and store on the deduplicating backup appliance at the home office.
This is the architecture that Quantum has adopted for its DXi3500 and DXi5500 appliances and its scalable DXi7500 system to facilitate the deduplication, replication and optimal storage of backup data across an enterprise. In the next blog entry, I'll take a look at how Quantum's DXi-Series supports this architecture and what configuration options this platform provides.
]]>Barry was recently put in the situation where he began to doubt that his data protection software was providing the level of protection that Blessing Hospital needed in order to recover its data. So before anything seriously bad happened to its data, he felt that a change in data protection software was required. However as he started to evaluate contending enterprise data protection software products, he could not ascertain which one was best suited for Blessing Hospital so he decided to bring in each of the contending products and test them.
When testing the products, he essentially put them through what he described as a drag race. Each product was loaded on the same server hardware and had to back up the same amount of data to disk. However what made Blessing Hospital's data unique is that it used a PACS (Picture Archiving and Communication System) system to create and manage its imaging files. The PACS system creates a large number of files that only change infrequently but need to be kept online for ready access to meet specific HIPAA regulations.
A particular vexing problem he ran across when backing up files on the PACS system is that periodically the PACS management database would touch a large number of files. Though none of the content in any of the files would change, the PACS system would in some way change permissions on the files. What he discovered in his testing is that the different backup software packages would interpret these changed file permissions differently. Some products would conclude that the files had changed and needed to be backed up again. This would dramatically increase backup times and the amount of data backed up.
What Barry specifically found noteworthy was how much the different backup software products differed in how they performed the file scan. In most cases, the larger the server, the larger the file system it had; the larger the file system, the longer it took to scan. In this aspect, the CommVault® Simpana® software suite distanced itself from the competitors in his environment. "The scan up front is where the magic is," says Barry.
Barry came to understand that CommVault's Simpana software is so effective in scanning files during backups because of how its intelligent agents interact with CommVault's underlying Single Instance Store (SIS) database. Its intelligent agents don't just look at a file's permissions but at the file's make-up. As the agent scans the file, if it determines the file permissions have changed but the file has not, CommVault knows it does not need to backup the file again. This, says Barry, resulted in backup times that were about half of what he achieved using Symanctec's Backup Exec.
Barry said the choice eventually came down to EMC Networker and the CommVault Simpana Suite. Even though he considers Blessing Hospital an EMC shop and has a lot of respect for EMC, he chose CommVault. Since his implementation of CommVault software, he finds that he has reduced the total amount of storage he stores on the back-end as well as the time it takes to do backups.
Equally important is what Barry said as I was finishing up my call with him. While single instance stores and reduced backup times are great, data restoration is where the rubber meets the road. In this respect, he has seen restore times decrease from an hour to 15 - 20 minutes.
Barry's methodical approach to selecting the appropriate data protection product for Blessing Hospital deserves some accolades and not only because he choose the CommVault Simpana software. No matter how much is written about a product or how many case studies one reads, one never really knows for sure (even the vendor) until you put the product through the paces in your environment. As Barry found out, SIS was a feature that he may never have thought to look for in the literature or ask of the vendor, but through testing it ended up being the feature that separated the CommVault Simpana software from the pack.
]]>In case no one has noticed lately, the number of ways in which companies can configure disk-based storage systems to protect their data has multiplied significantly. This fact was brought clearly into focus by a pre-recorded video lecture that I recently watched on Overland Storage's Tiered Data Protection (TDP) website. Though I think most users are well aware that disk is now a viable target for backup, it didn't really hit me until I watched this video by Overland Storage's VP of Worldwide Sales, Bob Farkaly, just how many configuration options are available when using disk as a backup target.
During the presentation, "Professor" Farkaly took the viewer (in this case, me) through the disk-based offerings that Overland Storage has for backup. These include:
It seems that with the incessant noise one hears about deduplication every time you read or hear about something storage related, it's easy to forget about the fact that using disk-as-disk is not necessarily a bad thing. Deduplication may incur a write penalty when backing up data, a read penalty when retrieving data and, more than likely, an up-front financial penalty when purchasing and licensing the deduplication technology. Though Overland Storage offers a deduplicating backup appliance (which I'll get to in a second), these versions of the REO peel away some of the complexity and cost and give users options to present the systems as either disk or virtual tape libraries (VTLs) through iSCSI and FC interfaces.The videos on Overland Storage's TDP website are definitely worth the time to watch for those who are looking to learn more about tiered data protection as well as those who also want a better understanding of what is the best type of disk system to introduce into your backup process. What you read in the press can sometimes leave one with the impression that deduplicating all of your backup data is the only way to go. Longer term it probably is, but for now the journey of introducing disk into the backup process has just begun and Overland Storage's strategy of "Walk, then run" when introducing disk into the backup process is one worth following.
]]>