Speaking at the AFCOM Souther California Chapter re: CMDB

Darrell Gardner will be presenting "The Poor Man's CMDB" at the Souther California AFCOM Chapter meeting on May 26th at 10:00 am.

Courtyard Burbank:
2100 Empire Ave
Burbank, CA 91594

Hope to see you there.

AFCOM Final- Evaluating Your Data Center Options

I was able to sit down with a number of AFCOM attendees that are looking for some direction in how to objectively evaluate the data center sourcing options for their Enterprise computing needs. They have a good idea of their datacenter space, power and cooling needs and even have preferences on specific critical facility design concepts, locations and sourcing scenarios, but seem to struggle with objectively evaluating these alternatives in relationship to their business need.
What is required is a solid understanding of the spatial and overall performance objectives for the facility and then be able to develop specific qualitative and quantitative criteria to evaluate the DC location alternatives.
The major Qualitative criteria are usually focused on regional disaster risks, network diversity & latency and efficiency potential (climate, free cooling). I like to include a broader set of operational criteria such as; vendor presence and proximity, IT staff availability, adjacencies to core business operation, utility power generation sourcing and general tax and regulatory incentives. You can weight the criteria relative to your specific needs and may want to use one or more criteria as Gates (such as regional disaster risk, or Facility reliability tier level). The idea is to distinguish and rate the quality of the data center location alternatives to one another. This can be done with specific properties or service provider sites or simply based upon metro area/geography.
The Quantitative criteria are simple. They are Costs. Develop a multi-year estimate of the costs associated the each sourcing alternative. These will vary with the sourcing scenario (owner operated/ co-location), but will usually include: the annual costs for facility or tenant improvement design/build (depreciated over time), Utility power, lease/rent/Co-location fees, predictive and preventive maintenance, IT operations staff/services and relevant network costs. Develop an estimate for each location scenario. This gives you a common Total Cost of Operation for each of the sourcing alternatives you are considering.
By combining both views of the deployment alternatives you can obtain a solid basis of comparison that takes into account the key operational and financial considerations required to develop a data center sourcing strategy.

What good is CMDB if it does not Tie into Change Management

No matter how cool the reporting you get from your CMDB tool is, if the information contained in it is stale outdated, or just plain wrong.. You are hosed.

The content of your CMDB is as valuable as the last person that updated it. Most likely the CMDB will be around longer than that person so make sure it is up to date.

Make sure that you update your CMDB as part of your Change process!!! Adds/Moves/Changes to any part of the data center should be required to use Change process. Part of the process should be a check piont that updates any dependent documentation. If you have a CAB board that reviews changes, the udpates shoule be required to be implemented as of a time limit within the change window or even as part of the change plan. Possibly the clean up prortion of the plan right after updating monitoring tools. Then review to verify that changes were put in place. Sign OFF!

Do this and you will be ready for changes to your data center environment. Especially in a hosted environment when you have hands and eyes that are less familiar with your environment (New Hires, Contractors, Co-Lo's).

Any Change to the environment or dependency -> Implement Change -> Update CMDB

Tying your Inventory/Assests back to Applications

As I have done many Data Center Migrations, I have found one common thread on the initial engagements....

There is a disparity amongst the support teams and the application owners as to what they call their applications. Support folks tend to shorten or give acronyms to systems and the application teams have their own names as well. This causes some problems within the change management arena as well as documentation. My suggestions are as follows:

• Standardize on the names of systems within your data center
• Standardize on all potential classification area's (ie. locations, support groups, Operating systems, function, etc...)
• Verify that every device ties back to a common system name.
• Within your system documentation you could then track:
•  System Name - Common or Highest Level System Name
• Component: Module or Subsystem Name associated with System (Blank ok)
• Function : Type of device (Websvr, DBSvr, Storage, AppSvr, citrix svr)
• Servicing Location: What locations or Enterprise does this device support?

By doing this you are on your way to having a healthy change management system with accurate information regarding systems and their associated owners, stakeholders, and other upstream or downstream devices/systems.

This is also imperitive in order to have a proper dependency matrix (I'll cover that in another post).

AFCOM Data Center World Day 1

With 800 attendees and 190 exhibits the conference got moving quickly with keynote, presentations and the first of two exhibit hall sessions. As the focus of AFCOM is data center physical infrastructure, the majority of power delivery, cooling and cabinet manufactures are well represented. I haven't run into anything revolutionary , but a few cooling/cabinet vendors have some interesting new lines. A few people who dropped by reported that a vendor had a running server submerged in cooling fluid, but I couldn't verify or track it down.

I met with Mission Critical publications and discussed the merits of managing utility power as part of a company supply chain and HPs smaller form-factor (20ft) DC containers. I think the smaller units can get some traction, particularly with IT executives who see it aligning with their own server technology deployment life-cycle. My discussion with Data Center News covered the challenges of bringing IT and Facilities together in DC initiatives and what kind of sourcing choices we see enterprise customers making today.

I was encouraged by the number of attendees today, but by the end of the day, it seemed only marginally improved from last year. We spoke with dozens of folks regarding their DC initiatives and challenges and was a bit surprised at the number of active DC deployment projects. There was also a couple of new entries for the 'best state to build your data center' competition with Missouri and S. Dakota economic development agencies exhibiting this year. Very few IT providers present with HP the only tier -1 provider and they had a relatively small footprint.

Data Center World- Expectations

It has been a pivotal year for many of our clients and I am interested in how the data center community represented here at Data Center World are responding to the challenges of data center design, sourcing and operation. If I had one word could describe the data center space today it might be "OPTIONS". There are so many choices of power and cooling technology, facilities architecture, location alternatives, collocation models, computing services and outsource providers. I am particularly interested in real-world green data center deployments for the typical enterprise user-operator, modular data center designs as well as any serious use of commodity computing/cloud services.

Last spring's Las Vegas conference attendance seemed impacted by the 2008/9 economy tsunami. When I arrived at the convention center today I ran into a number of folks who skipped last year’s conference, but are back this year. Seems like a good sign...