Wednesday, 8 February 2012

Vendor Infrastructure Stacks - do they deliver on their promise??

So another blog article on the topic of vertical stacks / engineered systems - however this time I would be interested in what the general views are out there... Of course I would never write anything without stating my views / opinions etc. ;-)

All these stacks / engineered systems / appliances are emerging at a fast rate - but I have to ask - really how useable are they and where is the real market place / market sector that they have real benefit.

Working in a number of large enterprises over the years - some of the big technology departments understand all of the pitfalls of putting stuff together wrong, ensuring that inter-op matrices need to be considered and also it doesn't just stop at the infrastructure layer - it also has to take into account the layered applications that sit on top (database, middleware, message buses, java runtime engines etc...). Why do we need to get this right, well (a) to ensure stuff works properly (b) ensure that we are using a supported configuration and finally (c) as a cover ass to ensure when something does go wrong - we get the support from any given vendor when we need to pick up the pieces and make it all work again!

So what is the promise of these so called engineered stacks, well my view is it should be some of the following:
   Ready to rock and roll
   Integration tested
   Ready to “just use”
   All that time making stuff works is taken away from a business (they can concentrate on making money rather than making technology work
   One throat to choke for support / outage issues
   Ease of upgrade between generations of stacks
   Simplicity of support
   Can consume shared services within a given organization to ensure that security, monitoring, authentication etc. capabilities can be available
   No need to engineer based on components – just validate functionality and you are I business

Of course - if you take a look at these stacks, they are also not made equal. Technologies such as VCE vBlock really deliver the core infrastructure layer where as vendors such as oracle provide the complete stack from infrastructure right up to an application layer (i.e. database / middleware etc.)

I also accept that different sizes of organizations may require different stories & capabilities for a stack… For the larger enterprise they may have comprehensive IT functions and staffing that allow full integration testing and validation (lets call it engineering capability) and others smaller enterprises may just want to consume technology without large I.T. functions… I also buy that the business wants to concentrate on just that (i.e. the business that they are in / making money rather than technology work for technology sake).

So back to the promise of these pre-engineered systems that we wish to consume for the above reasons – it really requires that the stacks are of a mature enough state that they can just slot into a business and be used… and here is where my problem really lies…. My personal view is that the level of maturity is still not there, and if you are not careful you can take on some of these technologies and turn it into a massive professional services gig (which is fine as long as you get efficiencies at the end) and potentially some increase in technical complexity (really should be a reduction!)

Some of the traits that I have read about / noted / observed would be:

·      Masses of professional services to implement a solution
·    Varying degrees of integration completed
·      Never quite the finished article
·      Management integration is poor
·      Internal engineering / technology departments still required to ensure that integration and take on activities can be handled
·      No consideration for organization that will be taking on infrastructure
·      The vendor offering stacks has a less clear view of how to do this integration than some of the technology depts. within enterprises (as those technology departments have probably been doing it longer)
·      A patch regime that was worse than before
·      Slide ware promises the world, actual product is left wanting (i.e. senior management go for the dream / vision but in reality product doesn't match slide ware)
·      Vendor Lock-in
·      Escalating costs
·      Lack of control
·      Destruction of the world
·      Wars that encompass the length and breadth of the universe

OK – maybe the last two are not quite fair – and I having a giggle – however I still believe there are maturity issues…

Enough of my ramblings and ranting’s… over to the reader!!!

So – I guess this is really where I want to put a question out there – which of you have used vendor supplied holistic stacks, how have you found them and to be explicit I would be interested to read / hear about some of the following points (snail mail / email / twitter etc. is fine)

·      What is the relative size of your org
·      Which vendor stacks have you looked at
·      Which vendor stacks have you chosen
·      What was the level of success
·      What perceived benefit have you found (and how did you measure it)
·      How much integration did you end up doing as the stack vendor had not completed all aspects
·      Would you do it again
·      What didn't work and why

I would love feedback in anyway I can get it and am happy to be challenged – so please feel free to use my blog page to respond – or all other forms of communication


1 comment:

  1. Hi Stu

    Dirk here...

    The challenge remains the same in my view:

    These stacks should align to a given service or set of services.
    You need to consider the total cost of a given stack (selection, purchase, commission, integration, support, refresh) and compare that cost with the alternative "internal build" approach to inform the decision.
    Some of the traditional CoLo vendors that are getting into the hosted private/public cloud market are doing a lot of the work for you, is it more prudent to consume their private cloud than build your own? Obviously that depends on the business, data centre ownership, capacity etc.
    The VDI stacks are beginning to look much more interesting than stacks for other services (e.g. databases, mail, VSI). But a lot of that has to do with the software eco-system.

    Another formal assessment ;-)