βTechnology Consulting Group Logica Australia

Last year several of us attended presentations ran by Opalis both in Sydney and Melbourne and called “Tech Workshop:  Operational Readiness for VMware and/or ITIL”. Having some time on the hands another chat compilation has been created:

Kelley Johnston: I thought this workshop was going to just be another VMWare product conference where a new version of VMWare was to be announced.  It was, and somewhere around election day a nice bit of VMWare VS-O product will surface.  Cool, but honestly not earth-shaking.  Essentially  (a) You will soon be able to boot VMWare hypervisor on a blank system from a thumb drive, and (b) Dell, IBM and HP will now let you buy servers with the VMWare Hypervisor in firmware, on the motherboard, supported by new chipsets from Intel.  Worth seeing for the free orange juice, maybe.However, in my opinion VMWare was completely upstaged by a product called Opalis Integration Server that implements what Gartner calls “RBA” (Run Book Automation) with a very comprehensive end-to-end data centre workflow orchestration tool.  It doesn’t just document or design work, it does work, as in actually scripts and runs the tasks for you.This is a graphic tool that allows you to develop workflows in the data centre, and it rides on top of some of our old favorites like HP Openview, BMC Patrol, Tivoli, MS SMS, Legato, Symantec, FTP etc. (many etc’s).  You grab an icon, draw connectors (interface objects), assign servers (or clone a VMWare instance) and other parameters, move files or trigger other events all via a graphic icon-based UI  full of nifty icons that click into object interfaces into products commonly associated with ITIL processes.  You can also define your own interface to anything that supports SOAP/WSDL web services.

What does this mean? 

It means you can define repeatable ITIL processes and automate them, cut down DC operational costs and improve quality.  I mean it.

In just one example, an operational event such as a Tivoli or HP Openview incident can be grabbed by Opalis which can do stuff with it — send emails, configure a new VMWare server after grabbing a configuration from a CMDB, perhaps swap VMX instances, do specific file copies via FTP, do a backup, manage and perform a recovery, add an AD user, run a cron job on some other Solaris server, keep a BMC Patrol support desk incident up to date, acknowledge an Openview event, reset a server, reset a user password, etc etc. and all within a single named RBA process.  And you can have processes call other processes, like little reusable ITIL scriptlets.

It’s almost like having a graphic programming language for all those hundreds of little things you do in a data centre.  You can also use it as an engine for defining a complex cross-platform workflow that describes say, 90% of a supply chain integration workflow using pre-existing components.  I suspect this product will grow beyond the data centre myself — check with me in a year on this.

I’d give the product at least 4 out of 5 stars just from the demo (even allowing for demo deviltry).  Very much worth a look.  Here’s a nifty screen shot to help, just below.    http://www.opalis.com .

Justin Hobson: Seems very interesting, but I just remembered that Opsware is supposed to be a global standard for this kind of toolset.

Either tool aside, these tools are usually great for automating processes, reducing the number of errors when it’s done manually and provided a definable and documented process.

On the flip side though, they are usually costly to implement and as such the most value comes from when they can be deployed in a traditional outsourced environment. e.g. with a structured network that hierarchically connects all out clients back to a central “Manager of Manager (MoM)” type networks, from where all our clients can be managed, with fewer engineers/servers etc.

Without such a backbone we would otherwise end up seeing many disconnected instances of these tools, that will mean increased complexity, licensing costs etc, and whilst bringing some benefits to us/our clients.  

http://www.opsware.com

Kelley Johnston: It’s a very interesting field.  Opsware is good too (incidentally it was acquired by HP in July this year).  Would be interesting to see a shootout of Opsware vs Opalis.  The latter product is considerably newer, looked very easy to use, very visual and coherent approach, and is apparently gaining market share as well as having being voted “Best of TechEd 2007″. So it’s gaining street cred, even if it doesn’t have the track record of Opsware.

This raises the question — do we keep with the standard or do we evaluate this upstart to see if it does a better job at managing DC operational costs?  Standards are good, but you have to weigh a product on the same scales our customers would use, and a better / newer product sometimes comes along that can achieve a better ROI.  We’re in this for our customers, not our selves, and they pay the bills.  We have to continuously look to improve our offers or our customers will do end-arounds on us.

Technology moves awfully fast.  In my experience “easier” generally equates to “cheaper”, so I wonder if the product has improvements that address some of the concerns you mention (I’m not an Opsware guru). As you say, though, either product has got to be better than chasing ops processes by hand.  An alternative product might also be attractive to customers who are aligned differently from the HP camp, too. I’d say the best advice is “watch this space”, because Gartner seems to have RBA on its radar now and they’re um, reasonably influential.

PS: http://www.byteandswitch.com/document.asp?doc_id=116542

Turns out the CEO of Opalis (Todd DeLaughter) is ex HP GM of the OpenView product line. 

Ex HP exec jumps ship to chair a product firm, HP buys competing product.