Records Management in the Cloud: Another perspective

Community Topic(s):

Keywords: virutalization, records management, cloud, future, ecm

Current Rating:
(0 ratings)

The technology world is certainly cloudy.  A part of that is a severe confusion of the terms used to describe “The Cloud.”  So before I say what I want to say, let me add some clarity.  There are three primary uses of the Cloud, and no it’s not to edit photos like Microsoft’s commercials would have you believe. First are Cloud file systems, second Cloud based applications (a.k.a software as a service SaaS), and third virtualization.

Cloud file systems, in the ECM world, are the most popular topic.  So when an ECMer uses the term, “The Cloud”, they are probably talking about solutions like Box.net or Dopbox which are Cloud file systems.

To the consumer, “The Cloud” is predominantly Apps, where all settings, and processing are happing somewhere else besides their own desktop/laptop.

However, if you talk about “The Cloud” with system administrators or IT managers, they most likely are thinking about virtualization.  Taking whole environments and putting them in “The Cloud” along with all the cool administration tools that come with it.

I’ve seen all of these worlds now, and found it’s important to understand which you are talking about.  I have spoken about the use of records management in “The Cloud”, and at that time I was referring to securing documents in a Cloud file system.  However, today, I want to talk about a slightly different perspective, and that is records management in the last category, the virtualization world.

At the core of “The Cloud” is virtualization.  This is the technology that makes the Cloud scalable, and less expensive than in the good old days, of dealing with mounds of physical hardware.  At the heart of virtualization is an entity called the Virtual Machine or VM.   A VM is the solitary representation of an environment’s, operating system (OS), software, and all configurations.  It’s like taking your whole desktop and putting it into a single file.  Just like any other file, as long as the right “viewer”, in this case a hyper visor, is available, it can be opened and run.  The problem is these files are very large, on average even a compressed VM is 20GB.

Other than its size, a VM is no different than any other file. Often shared as formats .vmdk or .ova, they have content ( OS, software, configurations ), meta-data, description, version, last modified date etc..  Here is where it gets fun.

In the not too distant future, are we going to have to worry about records management for VMs!?  I suspect yes.  At CloudShare, we already do, we actually have set retention periods for VMs that have been removed from an old environment or snapshot.  But at CloudShare we are thinking of retention more from an overall performance perspective, not a legal one.  What happens when legal gets involved?  What happens when whole environments are considered records in a case?  Could a judge ask for whole VMs to be turned over just like files? 

It’s actually already happening.  In the JVM world you have Oracle going after Google for their configuration of a Java based VM at the core of Android devices.  And in the world of software piracy, where the law is broken within a VM, handing them over to authorities is no big surprise.

I suspect that when the technical world is predominantly in “The Cloud”, thus in some VM somewhere, control over the VM files is going to be essential.  There will need to be a process to version whole VMs and VM snapshots, a process of record declaration, a process of destruction and retention, and even eDiscovery.

The war to control the virtualization is very heavy right now.  I call it the battle of hyper visors, the biggest contenders in the Windows space are VMWare and of course Microsoft.  The good news is that anymore it’s possible to convert a VMware VMDK file to a Microsoft Hyper-V environment.  Will it eventually just be a standard like PDF?  Probably.

The only hurdle preventing VMs from being passed around like word documents is their size, and the world’s exposure to them.  Both are solved with time.  High performance networks transferring data at gigabytes per second versus megabytes already exist, and will soon hit the masses.  So do compression and acceleration technology making files smaller and more portable.  These will make passing VMs effortless.  Additionally more and more people in the legal, records management, and judicial space will understand what a VM is, and demand some control around them.  So records managers beware, soon you will be adding VMDK files to your retention schedule.

Report

Rate Post

You need to log in to rate blog posts. Click here to login.

Add a Comment

You need to log in to post messages. Click here to login.

Comments

Greg Cook

Wow...

This is an area of RM I had not thought about before. It really got the creative juices flowing this afternoon. I found my brain racing around thinking of the possibilities and how to manage various VMs and what state each is in currently.

Report
Was this helpful? Yes No
Reply
Chris Riley, ECMp, IOAp

Declaring a VM a record

It is interesting. The aspect I've thought about the most is declaring a VM a record. Maybe you don't do it with living VMs and only snapshots. The issue is that to declare a VM a record you not only have to restrict the file, but restrict the VM when it's run. meaning you have to disallow any new content to be added to the client OS which includes installing software. You need the VM in a permanent state. Today the only way to do that is to have Snapshots, but not on living VMs.
Report
Was this helpful? Yes No
Reply
Greg Cook

Living VMs

That is a good point about living VMs. How would you handle something like routine patching? A living VM being declared a record could be a problem because each month pathes to the OS and underlying applications would give an RM fits.

But, declaring each snapshot a record and determining when to take a snapshot would be more managable. The disk space usage becomes mind boggling when you start to go down this path though. I might take a half dozen snapshots each month, unless I can bundle all my patches together for a single installation window.

A de-duplication storage solution would be mandatory at this point, just to ensure you don't end up with hundreds of TBs in snapshots.
Report
Was this helpful? Yes No
Reply
Chris Riley, ECMp, IOAp

Size

Well size of snapshots is interesting. Depending on the hyper visor there are several ways to take snapshots, and many include technology where you don't have to take the WHOLE environment. So a snapshot is large, but they are giving more and more options to keep them far smaller then the VM itself. Which is also interesting when it comes to record deceleration. What type of snapshot do you require?
Report
Was this helpful? Yes No
Reply

John Holliday

Journaling Required

Interesting idea, but thinking practically, it would likely require journaling support so that all changes to the VM over a span of time could be recorded and selectively played back. Static snapshots would not, IMO, capture enough information to be useful, especially when defending legal challenges.
Report
1 people found this helpful, did you? Yes No
Reply
Chris Riley, ECMp, IOAp

Only for Live VMs

Interesting point. And this is available with many Hyper Visors.
Report
Was this helpful? Yes No
Reply

Silke Dannemann, ECMm BPMp

Journaling and then some required

To capture more, you would then need to turn to open (preferred) or proprietary standards to capture the deltas.
Report
Was this helpful? Yes No
Reply

This post and comment(s) reflect the personal perspectives of community members, and not necessarily those of their employers or of AIIM International