Provoking IT from Good to Great
The Z-Factor: Scaling deep in the data center
Jay-Z - VMware Certified Professional
We’ve had X-Factor scale-out; we’ve had Y-Factor scale up: now we’ve got Z-Factor scale deep.
If you’re using virtualization in the data center, then you have the Z-Factor, but how big is yours?
The Z-Factor is a measure of your Efficiency and Effectiveness of your data center <- don’t for one minute think that this is boring, here’s why:
Efficiency is the goal of your IT Operations to deliver best-of-breed services at the best cost – the old price/performance ratio. If you’re bad at this, then the Cloud is going to eat your lunch.
Effectiveness is the goal of your Application and Business teams who leverage predictable (availability) and responsive (time-to-market) IT to be more competitive. If you’re bad at this, then the Cloud is going to eat your dinner.
Steve Duplessie at ESG recently wrote about IT Mis-alignment and in a white paper made a great point about the different goals of the IT operations group vs. their customers. If you’ve been in virtualization a while you will have had or heard of the Virtualization Pitch by IT Ops to the Apps guys, and it goes something like this:
IT Guy to App Team: We want to run your new stuff on virtual machines because it means we don’t have to buy you real computers, which saves US lots of cash in reduced power, cooling and hardware.
App Team to IT Guy: Screw you! You’re offering me an EVEN WORSE service! What’s in it for me?!
If you don’t believe me, watch this slideshow 10+ Deploys Per Day: Dev and Ops Cooperation at Flickr.
The Z-Factor Explained
Quick refresh:
- X-Factor Scale-out means buying more servers.
- Y-Factor Scale-up means buying more compute per server.
- Z-Factor Scale-deep means virtualizing the network, storage and compute to dig deep and return more value from your investment.
The Z-Factor is a measure of Return on Investment (ROI) not Total Cost of Ownership (TCO). The better your Z-Factor, the better your ROI. And this is measured in two parts: your Density and your Response.
Density
How much is IT Ops exploting virtualization?
Virtualization is more than just compute virtualization. You can virtualize your storage. You can virtualize your network. Net:net, the more virtualization components you have in play then the more you are “sweating your assets”.
It’s not enough just to architect for a specific ratio (e.g. 50:1 Guest:Host), what we want to see here is evidence of active operational capacity planning and workload lifecycles that recycle capacity back into the system.
The ESG report put consolidation as the #4 priority of IT Ops, when really it’s not a priority but an enabler. The friend of consolidation is containment – having a Virtual First Policy to ensure you are reducing the future sprawl of physical machines.
Examples of Density
- Guests:Cluster Compute ratios are more important than Guests:Host
- Using Network virtualization through Converged Network Adaptors, Fibre Channel over Ethernet (FCoE) and Nexus combined-function switches.
- Storage wizadry like Thin Provisioning.
Response
How much are the Applications/Developers/Business exploiting virtualization?
If you were in charge of developing and releasing a new car to market, then you have a bunch of constraints on you: time and money chief amongst them.
To get your car to market, you need IT resources. If someone from IT says that they want to run your stuff on virtual machines, which might run a bit slower or a bit faster, or cost a bit less or a bit more, who knows: what would you say?
As the guy responsible for getting a product to market your requirements to IT are simple:
- Help me understand what IT resource I need: what are my requirements?
- Satisfy my requirements in a timely manner, preferably in anticipation of demand.
- Our requirements will change because product development is unpredictable: so IT be ready.
With virtualization you have the ability to flexibly respond to these changing needs: this is the other side of the Z-Factor.
Examples of Response
- Increase the processing capabilities of workloads without having to rebuild the system – just add CPUs, Memory, Disk, Network via vCenter and UCSM.
- Reduce and re-assign the processing capabilities as the projects move through phases.
- Provide a self-service portal to the developers so they can get on with their work in a dynamic manner and not have to route via the IT team: this is transparent IT.
Summary
Have you got the Z-Factor? It is a measure of Density and of Response.
If you are in the data center end of IT, then you are responsible for deploying best-of-breed at right-cost infrastructure that is leveraging all facets of virtualization across compute, network and storage to increase the Z-Factor. The more efficient (dense) your infrastructure, the better your return on investment.
If you are in the application / business end of IT, then you are consuming IT in a way that maximizes your productivity at a reasonable cost, in a way that provides your organization competitive advantage. The more effective (responsive) your infrastructure, the better your return on investment.
Here’s the sales bit: You need VMware vSphere, Cisco UCS, and a management framework to achieve the Z-Factor
Related posts:
| Print article | This entry was posted by Steve Chambers on 23 August, 2009 at 22:49, and is filed under Operations, UCS, good2great, vSphere. Follow any responses to this post through RSS 2.0. You can leave a response or trackback from your own site. |
about 1 year ago
Once again you bring up something I’ve been sensing in the IT trenches.
The quote just below Jay-z’s picture is exactly what I hear everytime a new project shows up on the radar.
Interesting thing is, the developers/app team never seem to want to build code in an efficient manner first. They always seem to ‘throw hardware’ at the problem until the code develops into something that they can use. A generalization for sure, but the efficiency of app development isn’t measured by TCO or ROI as far as I have seen.
Case in point is the 3 tiers of systems designed for the development of applications; ie. Development, Integration, Certification/QA . Does it matter how many systems / environments can be created for the App teams to be satisfied in their hunt for optimized code?
If we took all three tiers of that environment for an application and divided it by the development time of the application from design to production would it calculate to a number that the business would be willing to swallow? More importantly did it match the High Level Estimate at the start of the project.
Don’t get me wrong, virtualizaton certainly helps the Z-Factor tremendously. I just don’t think the application teams can deliver at the levels they will require if they perceive that virtualization degrades their service. They should instead see what their code really requires (at the end), instead of what they expect (in the beginning).
about 1 year ago
A brief but brilliant summary of the issues. I feel that a POC from VMWare and Cisco which will have self-provisioning of virtual servers, and an out-of-the-box way to create clouds that are compatible with Amazon, Azure and GoogleApp may be the way to lift the fog, and drag the Cloud into the DataCenter or the Datacenter into the Cloud.
Yes, Cisco UCS and VMWare are key to this.