There is no major issue with the concept of a central database which holds details of every instance of an organisation’s servers, operating systems, hypervisors, network equipment and end-user applications, showing what is used where and the dependencies between them. That explains why ITIL promotes configuration management databases (CMDBs).
The trouble is that this takes a lot of work, some of it manual, to achieve an accurate and up-to-date CMDB and, equally important, to keep it so—so that it is really useful. In truth this is a long-term project and most businesses are not yet very far down that road.
Much of the information a CMDB will contain is still needed in order to properly manage the infrastructure and keep control of software, for instance to comply with licensing requirements without being overcharged. To stay up-to-date means keeping on top of every change that occurs. Nowadays this includes identifying when virtual servers and their contents are created or collapsed.
The only real solution is automated discovery—but even then it needs to be comprehensive and accurate. Here is where quite a few major vendor offerings are lacking.
London-based Tideway Systems has invested a lot of time and resource in getting discovery right to create its Foundation discovery software. The company has the advantage of being entirely independent of the large hardware suppliers; these, as you would expect, tend to provide a more comprehensive picture of their own equipment and software than that of competitive vendors. Conversely, a smaller privately-owned player needs to offer something extra to really compete.
The first point is that its discovery software is agent-less. According to Kosten Metreweli, Tideway's VP of product marketing, using agents will tend to pick up less than 70% of the devices on the network (as well as there being typically 10–13 different agents in a datacentre). However, the discovery has to be deep and Tideway expects its discovery to turn up some previously unknown systems when the software is first applied.
Equally important is enrichment of the bare item information, so the company leverages a trusted third party source for reconciliation. This is not unusual and is not always enough. "Our approach," said Metreweli, is for the software to work in the way a human-being would recognise [for instance] an Oracle database."
Put another way, it applies probabilities when given incomplete data which leads to further checks—a flexible approach similar in principle to what the CMDB-creators have to do manually as they work through the information they receive to try and reconcile it.
Yet it seems to me the toughest task—but potentially the most useful—is to identify all the dependencies between all software and all hardware, including use of virtual servers and their applications. Tideway has clearly cracked this. One of its major financial customers is a large global bank which, as Metreweli told me, has some 40,000 servers, thousands of business applications and millions of dependencies. This can be a nightmare for IT managers but is by no means unique among large enterprises.
The Foundation software can plot the lot, and can if required produce a Visio-style map of the whole infrastructure. However, unlike Visio, it is automatically formed from live up-to-date information—so you can show the boss what the true picture is like as at start of play the day you see him.
Browser-based tools that navigate the information will drill down to any level to help in reconciling the non-discoverable stuff and produce graphs. But perhaps the most rewarding aspect is being able to answer the request: "Show me the number of dependencies" for a piece of software. This is hugely advantageous for finding out the precise licensing situation (unlicensed, or unused but licensed) with a potentially huge and rapid payback. (As such it will logically interface with an organisation's fixed asset register.)
These sorts of benefits are what will be expected from a CMDB—but only when properly set up. So surely the priority is to go for the best automated discovery you can find—as a pre-requisite to ever achieving a CMDB. It is for enterprises to check out the automated discovery products like Tideway's Foundation to see what best suits their needs.