Nick Toozs-Hobson

I refute CoTS as cheaper in Cloud

Blog Post created by Nick Toozs-Hobson on Mar 12, 2015

When  we have conversations with most of our customer CIO and CTO about Cloud, it is about how to transform to the “consumable” Data Centre. The biggest challenge we all see is traditional IT having to embrace the brave new world of DevOps (Shadow IT). Not only having to handle this explosion of semi-structured and unstructured data, but also managing access to, security & governance of data, workflows and the evolution of "hand-held" applications. Interestingly, they now have a triple headache;

 

  • They still have to manage data through its traditional lifecycles
  • They now have to present new and re-purposed dark data for re-interrogation.
  • They have to develop new (mobile) apps at the speed of light, for which traditional IT is not designed

 

This is leading to some massive challenges in change management. Consequently we are getting many requests to help to achieve new ways to manage IT. Software-defined everything is about the  simplification, automation and ease of use of systems for the end user. This should reduce many costs; administration and deployment in particular, but it does not necessarily reduce other costs. The concept of using Commodity of the Shelf (CoTS) hardware for all transactions and applications is the biggest fallacy of all. Installing CoTS "cheap and cheerful" servers and storage boxes, can actually cost you more. In our experience price does not equal cost! Two simple analogies:

 

  • the argument for LED lamps these days: it is a very expensive light bulb to buy usually 3 or 4 times the price of an old-fashioned bulb, but it is a bulb that consumes 60% to 70% less electricity and lasts for 10 or more years (6 or 7 times longer than traditional light bulbs) and it therefore actually COSTS you less even though the PRICE is higher.
  • smartphones or tablets, they are more expensive to buy than a PC, but they are much easier to use; they come with a working ecosystem that makes them easier to live with and they are portable. The thing is that they are built on specifically designed HW which is engineered to work best in that particular ecosystem so we are prepared to pay more for them; a secondhand Apple iPhone is still more expensive than a new Windows desktop on eBay!

      

This is also true of Cloud; systems need to be designed to “just work”. that means working with production requirements of security, up time, performance or recoverability. They need to match line of Business Service Level  agreements (SLA) and Quality of Service (QoS). This should all be managed at a Software layer to automate and mask administration complexity, but should run on the appropriate Hardware platforms behind it to deliver at the right overall cost. Everything must be fit for purpose in the architecture to deliver against the business requirement, and often "Cheap" is a lot more costly.

 

The principle of consumption as a service (use your mouse to configure, order, pay, complete and withdraw) must meet corporate performance, security, recoverability, auditability and reliability demands. And, as consumers, we shouldn't care how it gets to our desktop or mobile device provided it meets our specific Business requirements. But as Cloud providers we have to architect the most economical systems to achieve those goals, often with more cost effective components that are not the cheapest off the shelf.

 

This has never become clearer to me than in the last 6 months that we have been working with ServiceMesh (recently acquired by CSC). Their Software-defined management platform called Agility ingtegrates to the Software-defined orchcestration layer of UCP Director which accesses the power of the Software-defined Hitachi Virtualisation engine for storage and compute. This Software access the power of the hardware which enables a massive return in economies of:

  • effort
  • power consumption
  • floor space
  • time to deliver
  • time to tear down
  • time to reconfigure
  • consolidation
  • utilisation

 

The reason ServiceMesh chose to work with Hitachi is specifically for the hardware functionality in and the orchestration of UCP and its cost effectiveness in a Cloud environment.

 

Keep your ears open and your eyes peeled at Connect 2015 in las Vegas for more.

Outcomes