IT crapshoot: cost-cutting is costly in disaster recovery, archiving for e-discovery

Disaster recovery and archiving are key zones of interaction for IT and Legal Departments. When a lawsuit is filed and an e-discovery production request is received, a company must examine all of its electronically stored information to find documents that are relevant to that lawsuit. Court battles may arise regarding the comprehensiveness of the examination, the need to lock down potentially important documents and metadata, and the cost of identifying, collecting, preserving, and reviewing documents — all of which are related to the way in which data is stored.

Photo credit: Josie Hill

Photo credit: Josie Hill

With this in mind, I recently sought out Jishnu Mitra, President of Stratogent, a specialized application hosting and disaster recovery services provider, to obtain his perspective on disaster recovery best practices and the relationship between disaster recovery and e-discovery. Key points he made include:

  • effective disaster recovery sites are “hot” sites that can be used for secondary purposes rather than remaining idle;
  • “cold” sites are unlikely to get the job done and are not cheap;
  • efforts to keep IT budgets down by delaying or limiting disaster recovery, or by limiting archiving, can backfire;
  • budget-conscious IT departments are more likely to use archiving features built-in to their software of choice;
  • many IT and Legal personnel have a habit of being disrespectful towards one another and doing a poor job of communicating with one another;
  • more crossover Legal-IT people are needed.

Bruce: Can you provide a little background about Stratogent’s domain expertise?

Jishnu: We offer end-to-end application hosting services, including establishing the hosting requirements and architecture, hardware and software implementation, and proactive day-to-day application management including responding to any issues that arise. Most of the time we are tasked with building a full data center, not the building itself, of course, but a complete software and hardware hosting framework. We aren’t providers of any specific business application (like salesforce.com does). We design, deploy and operate all the layers on which modern business applications are hosted including the application’s framework e.g. .NET, Java or SAP Basis.

Our customers include multi-office companies, who require applications shared between offices, and web-based application SaaS (“software as a service”) companies. The scope is typically quite complex – we don’t build or manage general web sites or blogs — that’s a commodity market and too crowded. We build and manage custom application infrastructures for enterprises or for complex applications that require a range of IT skills to manage. Our customers hire us because they don’t want to budget to hire all of the people they would need to do this internally, or when they are deploying a new application that is beyond the current reach of their IT team. For example, if a company wants to start using a new-to-them ERP [“Enterprise Resource Planning”] application like SAP or (say) a Microsoft based enterprise landscape that needs to scale, we can multiplex our internal pool of talent to give their application 24-7 attention far cheaper than the company can hire and retain the specialized employees they need to do it themselves.

Bruce: So you supply the specialized competencies needed to build and operate complex application environments so that your customers can focus on their core competencies? Then their core competencies don’t need to include what you do in order for them to succeed.

Jishnu: Yes. They know what they want, they conceptualize what they want, but not the hardware they need and the infrastructure software. We can go in from the very beginning saying, “Here’s how you set up a highly available, clustered server farm for your social networking app,” and so on and so forth. We know how to customize it and set it up. They also don’t have our expertise in negotiating with hardware vendors, or in capacity planning, etc. Plus there’s the build phase, loading OSes, etc. We essentially give them over the course of our engagement the entire hosting framework on which the app runs and then take care of it for the long run.

Once we get their hosting framework to a steady state, they get to run with it for two, five or longer number of years with no or limited failure. So their role is conceptualizing on day 1 and then we become a partner organization worrying about how to realize that dream, handling inevitable IT break-fix issues and managing changes over the entire life span of that system. Disaster recovery usually becomes part of that framework at some point.

Bruce: Can you give me some broad idea of the scope of disaster recovery work that you do?

Jishnu: Disaster recovery is not a separate arm of our business. It’s very integral to the hosting services we provide. We build disaster recovery sites at different levels of complexity. It can go from a small customer up to a really large customer. And over time Stratogent gets into innovative approaches to deal with disaster recovery. The philosophy of Stratogent is that we’re not trying to sell a boxed solution to all the customers. It’s more of a custom solution, not a mass market product. We say we will architect and host your solution – and as architect we always add very specific elements for our customer, not just one solution for everyone.

The basic approach, even for small customers, is to choose a convenient and correct location for the disaster recovery site and use a replication strategy based on whatever they can afford or have tolerance to accept. As much as possible a disaster recovery site should be up and running and ready to go at a flip of the switch. They can use the excess capacity at their disaster recovery site at quarter end to run financial reports or for other business purposes, plus it can be used for application QA and staging systems. They can be smart about it, and keep it on, so that they can have confidence in it.

Of course a disaster recovery solution like this can’t be built in just a month or two – to do it right requires creativity and diligence. In one recent instance when asked to do it ”right now”, we had to go with a large vendor’s standard disaster recovery solution for our customer. Everyone knows that this does not get us anything beyond the checkmark for DR, so the plan is to go to a Stratogent solution over time, build a hot alternate site on the East coast, and sunset the large vendor’s standard disaster recovery arrangement.

Bruce: Given the importance of disaster recovery for a number of reasons, how seriously are companies taking it?

Jishnu: Everybody needs it, but it suffers from “high priority, low criticality”, and the problem rolls from budget year to budget year. Some unpleasant trigger like an outage, or an impending audit instigates furious activity in this direction, but then it goes on to the back burner again. In the recent instance, although disaster recovery was scheduled for a later phase for technical reasons, for SOX compliance the auditor demanded a disaster recovery solution by year end or our customer would fail their audit. So we went out and obtained a large vendor’s standard disaster recovery solution, which met the auditors’ requirements but isn’t comparable to a “hot” disaster recovery site.

The way disaster recovery solutions from some of the large vendors work is this: they have huge data centers where their customers can use equipment should a disaster happen. Customers pay a monthly fee for this privilege. When a disaster strikes, customers ship their backup tapes out there, fly their people out there, and start building a disaster recovery system from scratch. And by the way, if you have trouble here’s the menu for emergency support services for which they will charge you more. And in 95 out of 100 cases it just doesn’t work, but is a monumental failure when you need it most. These are “cold” sites that have to be built from the ground up. It takes maybe 72 hours to get them up, rather to be asserted as “up”. Then, as someone like yourself with application development experience knows, it takes weeks to debug and get everything working correctly. And when you’re not actually using them, standard disaster recovery services are charging you an incredibly high amount of money for nothing except the option of bringing your people and tapes to their center, and then good luck.

Bruce: You mentioned running quarterly financials, QA, and staging as valuable uses for the excess capacity of “hot” disaster recovery sites. Could this excess capacity also be used for running e-discovery processes when the company is responding to a document production request?

Jishnu: Possibly, but I haven’t seen it done yet in a comprehensive manner. The problem is you still need to have the storage capacity for e-discovery somewhere. The e-discovery stuff is a significant chunk of storage, maybe tier 2 or 3, which demands different storage anyway, so it makes sense to keep the e-discovery data in the primary data center because its easy and faster to copy, etc. That said, it is very useful to employ the capacity available in the secondary site for e-discovery support activities like restoring data to an alternate instance of your application and for running large queries without affecting the live production systems.

Bruce: Do you deploy disaster recovery solutions that protect desktop drive, laptop drive, or shared drive data?

Jishnu: As I have said, our disaster recovery solutions are part of whatever application frameworks we are hosting. We as a company don’t get into the desktop environment, the local LANs that the companies have. We leave that to local teams or whatever partner does classic managed services. We do data centers and hosted frameworks. We don’t have the expertise or organizational structure to have people traveling to local sites, answering desktop-related user queries, etc. But any time it leaves our customer’s office and goes to the internet, from the edge of the office on out it’s ours.

Bruce: But when archiving is part of the customer’s platform hosted by you, it gets incorporated in your disaster recovery solution?

Jishnu: Yes.

Bruce: Is Stratogent involved when your customers must respond to e-discovery and regulatory compliance information retrieval requests?

Jishnu: Yes. For example, we recently went through and did what needed be done when a particular customer asked for all the documents in response to a lawsuit. We brought in a consultant for that specific archiving system as well. Our administrators collaborated with the consultant and 2 people from the customer’s IT department. It took a couple of weeks to provide all the documents they asked for.

Bruce: Was the system designed from the outset with minimizing e-discovery costs in mind?

Jishnu: Unfortunately no. In this case archiving for e-discovery was an afterthought and was grafted on to the application later and a push-button experience wasn’t in the criteria when designing this particular system. But it woke us up. We realized this could get worse.

Bruce: So how do you do it differently now that you’ve had this experience?

Jishnu: Here we recommended to our customer that we upgrade to the newest version of the archiving solution and begin using untapped features that allow for a more push-button approach. Keep in mind that e-discovery products weren’t as popular or sophisticated as you see them now.

Bruce: Aren’t there third-party archiving solutions also?

Jishnu: There are several third-party products and you see the regular enterprise software vendors coming out with add-ons. We’re especially looking forward to the next version of Exchange from Microsoft, where for us the salient feature is archiving and retention. Only because email is the number one retrieval request. On most existing setups getting the information for a lawsuit or another purpose takes us through an antiquated process of restoring mail boxes from tape and loads of manual labor. It’s pretty painful, it takes an inordinate amount of time to find specific emails, its not online, it takes days. For this reason we’re looking forward to Exchange 2010 which has features built INTO the product itself. Yes, some other vendors have add-on products that do this also.

Bruce: And I assume you’re familiar with Mimosa, in the case of Exchange?

Jishnu: – Like Mimosa, yes. But when it’s built-in the customer is more likely to use it. By default customers don’t buy add-ons for budgetary reasons. It’s so much easier if the central product has what we need, and that is in fact happening a lot these days. I won’t be surprised if products in general evolve so that compliance and regulatory features get considered integral parts of the software and not someone else’s problem.

Bruce: Do you have other examples of document retrieval from backups or archives?

Jishnu: Actually there are three scenarios where we do document retrieval. Scenario one, which we discussed, is e-discovery. Scenario 2 is when we have seen retrieval requests in acquisitions, mergers and acquisitions, and we had to pretty much get information from all sorts of systems, a huge pain.

Scenario 3 is SaaS driven. For many of our customers, the bulk of their systems are either on-premises or hosted by Stratogent, but some of our customers use SalesForce.com or one of many, many small or industry specific SaaS vertical solutions. In one recent case, one of these niche vertical SaaS vendors, because of some of the issues in that industry, was about to go out of business. We had to go into emergency mode and create an on-premise mirror, actually more like a graveyard for the data, to keep it for the future, to enable us to fetch the data from that service. We figured out a solution for how to get all the customers’ data and replicate and keep it in our data center and continuously keep it up to date. Fortunately the vendors were cooperative and allowed access through their back door to allow us to achieve this. I call this “the SaaS fallback” scenario. SaaS is a great way to quickly get started on a new application, but BOY, if anything happens, or if you decide you aren’t happy, it becomes a data migration nightmare and worse than an on-premises solution because you have no idea how it’s being kept and have to figure out how to retrieve it through an API or some other means.

Bruce: In e-discovery and other legal-driven document recovery scenarios, how important is collaboration between IT and Legal personnel, or should I say, how significant a problem is the lack of this collaboration?

Jishnu: I’ve seen the divide between IT and legal quite often. Calling it a divide is actually being polite; at worst both parties seem to think the others are clueless or morons. It’s a huge, huge gap. And I have also seen it playing out not just in traditional IT outfits, but also product based companies when I was principal architect at Borland. When attorneys came to talk to engineering about IP issues, open source contracts or even patent issues, there was no realization among the techies that it was important. In fact legal issues were labeled “blockers” and the entire legal department was “the business prevention department”. And there is exactly the opposite feeling in the other camp with how engineering leaders don’t “get it” and how talking to anybody in development or IT was like talking to a wall. The psychological and cultural issues between IT and legal have been there for a while. In some of the companies that have surmounted this issue, the key seems to be having a bridge person or team acting as an interpreter to communicate and keep both sides sane. Some technical folks I know have moved on to play a distinctly legal role in their organizations and they play a pivotal role in closing the gap between legal and IT.

One Reply to “IT crapshoot: cost-cutting is costly in disaster recovery, archiving for e-discovery”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s