Foundation and the Data Center Baseline: Phase 0 of any big project

  by Simon Woodward | 25 Feb 2009 | Permalink |

The most recent sprint of work in Foundation’s engineering department focussed on making Foundation into the obvious tool for doing a datacenter baseline. I’m not going to wander off into a complete discussion of how our clients and partners go about doing baselines, but by way of introduction…

No large IT project, especially one that will affect hundreds or even thousands of servers, can be started without a baseline. Baselines for this purpose are also referred to as “IT audits”, and generally involve an in-depth investigation into the IT infrastructure to discover:

• Its extent: are there hosts that you’re not aware of? The answer is invariably “yes”. What are they doing?
• Inventory: What hardware and software are the hosts running? Are they physical or virtual? Where are they located?
• Dependencies: This is the longest, most complicated and expensive part of the baseline.
  • Which hosts and software provide which services?
  • Which IT staff own and operate the hosts?
  • Which business services (business applications) are provided by which infrastructure?
  • Which business units rely on which business applications?
  • What are the interdependencies between hosts, and therefore between business applications?

Virtualisation and consolidation projects also require an understanding of the performance of the infrastructure: how busy the hosts are, how much power and cooling they require and if there are any reasons why they should not be virtualised/consolidated.

The ITIL V3 definition of a baseline is:

A Benchmark used as a reference point. For example,

• An ITSM Baseline can be used as a starting point to measure the effect of a Service Improvement Plan
• A Performance Baseline can be used to measure changes in Performance over the lifetime of an IT Service
• A Configuration Management Baseline can be used to enable the IT Infrastructure to be restored to a known Configuration if a Change or Release fails

This is subtly different from what we need from the baseline. Rather than a reference point, what we’re looking for is more a state of understanding. And rather than focussing exclusively on configuration, the baseline needed by a broad-reaching IT project needs to focus more on dependencies – “if I change this, what’s likely to break?”

Where Tideway Foundation can help

Foundation gathers essential hardware and software information quickly and accurately.

Foundation’s Baseline Dashboard tells you the progress of your baseline at a glance: how much of the estate you have gathered information about, what the spread of operating systems is on the hosts discovered so far, how many are virtual, graphs of the progress you’re making and more.

Where Foundation really shines, however, is in helping you to determine the relationships and dependencies in your estate.

Client workshops, interviews and ad-hoc conversations with technical staff are the most common techniques for finding the business dependencies within and on your data center.

Before booking these people’s valuable time, you must know which questions to ask of which people. You have to have a lot of detailed starter information about how the estate is structured: which hosts are communicating with which others (with a rough idea of why), which hosts seem to be working closely together, which hosts seem to be relied on by a large proportion of the estate and so on.

Normally, this initial analysis is done manually by examining lists of processes, the network communications they’re involved in, the hosts’ communications… for many hundreds or even thousands of hosts, it’s a tedious, error-prone, exhausting exercise taking considerable time.

Once the initial analysis is complete, the workshops are conducted. These involve talking to on-site experts about your conclusions from the analysis. These conclusions will often be challenged: either your expert is aware of a subtlety that you hadn’t picked up on, or the environment has moved on a bit and they’re not aware of some new condition. Either way, you need a way to present your conclusions to them with clear supporting evidence that can be discussed and further analysed. Frantically cross-referencing reams of reports under the gaze of an impatient technical architect wastes time and precious good-will.

Foundation makes all of this a lot simpler:

Host Grouping automatically analyses the observed network communications between hosts to determine which hosts work closely together, which are depended on by large amounts of the estate, which consume the same services and so on. The estate is divided into logically associated groups after every discovery run, providing you (within minutes of your first scan) with a report that would take hundreds of hours to assemble manually.

Host Profiles can be generated for every discovered host. Each profile contains essential information about the host’s hardware and software, as well as pre-analysed tables and visualisations that give you an accessible view of what the hosts are doing. Print them, take them into meetings, discuss them and scribble on them – they’re an invaluable tool for collaborating in workshops and interviews.

The results of this first sprint represent our first piece of work to make Foundation into the tool of choice for organising and understanding all of your IT data. Watch this space: it’s only going to get cooler!