The Bloor Perspective: The Microsoft settlement, the NTL challenge, and SLAs

The settlement of the Microsoft case seems to have provoked strikingly different reactions from different sections of the IT world.

The settlement calls for:
- Windows APIs to be fully documented and published
- rival developers to be given XP code within 12 months and subsequent releases before their commercial launch
- Microsoft to desist from forcing PC hardware suppliers to pre-load Windows and from offering preferential licences for doing so and penalising suppliers who don't play ball
- Microsoft to pay for a three-person team to scrutinise its compliance on a daily basis.

In our view, there are two key issues. How Microsoft should be censured for past practices and how it should be prevented from repeating those in the future. Neither seems to have been addressed properly.

The 'tough rules and regulations' amount to this: Microsoft has to fully document and publish Windows APIs, which is arguably in its interest anyway with a monopoly already established. The problem though is that any APIs that aren't fully declared in the future would take some time to find and prove and would, of course, be "regrettable errors". And, anyway, you can have great fun screwing the competition through changing APIs, as IBM has demonstrated in the past.

Further to that, the API directive comes with a caveat that excludes cases that might "compromise the security of anti-privacy, anti-virus, licensing, DRM encryption and authentication systems".

Worse, there are no restrictions on future bundling of software or pricing, which is how the DoJ sorted out the IBM monopoly to all users' benefit 30 years ago. The only real hope seems to be, as I have suggested before, that the parallel and wider ranging EU investigation will not take too long and will have more backbone.

*Cable: high risk, high reward?*

It couldn't go on forever. Cable company NTL has been buying up its competitors for a couple of years now - making sure that it had the biggest possible share of the market without quite troubling monopoly investigators. Eventually, though, the time had to come when it ran out of competitors to buy or the money to do it. It seems to have been the latter.

The latest financial figures from the company show that it has lost over $4 per share in its third quarter - this means $1.03bn has gone in the last three months. A lot of this has come about as a revaluation of assets but NTL is also incurring interest charges that would match the total revenues of many large businesses. At more than $350m per quarter, NTL is paying pretty heavily for its acquisitive past.

While the company continues to grow its revenues and its customer base it is not succeeding quite as well as it should. It has only just broken through the one million barrier for the number of digital TV installations in the UK, and its up sell techniques don't seem to be doing too well with the average customer revenue rising by only 12 per cent.

NTL though is a company that is being stretched and this means is that customers sometimes aren't getting the service they require. It is clear a small investment in people might generate much greater returns for NTL. However, before it can do that, the company is looking to stabilise its finances and become profitable by 2003. In order to do this it has to shift $16bn in debt.

Ultimately though, NTL needs to improve its ability to deliver its products. There is enormous demand for broadband connectivity that will continue to grow. Add a basic diet of TV and telephone calls and the proposition is a good one. The trick now is to get them to deliver it!

*SLA checks and balances*

In the past IT users and providers have often attempted to create and monitor SLAs with technical terms, tools and measurements. But now they need to buckle down and develop SLAs that every end user can understand and that means including simple terms and measurements like the level of service, response times and such like.

This need to measure real application response times and availability from the users' perspective has created a demand for tools that address end-to-end tools monitoring. And there are a number of solutions cropping up each with their own advantages and disadvantages.

Agent-based systems, for instance, deploy small programs on the systems of end users to measure their use of the applications and measure the end-to-end response times of the transactions. The advantage of this method is that real transactions are measured on real user systems. The disadvantage is that the tool depends upon the user being at work and accessing the services to be measured.

Robot transaction simulation systems can be programmed to perform regardless of user activity and can provide good results even when user activity is low or performance is so slow that real users abandon the service. But few organisations will let robot simulations update their real data systems - so you often get skewed results.

Nonetheless, these tools must be used with discretion and it is essential that some basic questions be asked before SLA monitoring takes place. Chief amongst these is the absolute requirement to identify what services are important to users and to ensure that reasonable, measurable SLAs are agreed.

With this knowledge, appropriate technology tools can be put in place to supply the measurements that qualify the quality of the service that is delivered. Some of the companies that can supply SLA monitoring tools include CA and HP who have agent-based technologies within their enterprise management offerings. Fujitsu Softek has a transaction simulation robot based tool in their EnView product while Candle ETEWatch and Tivoli APM supply alternative tools.

It is clear that technology alone is not enough to guarantee good service delivery. With enterprises now linking with the systems of their partners and suppliers along with providing customers with direct access to their systems it is essential effective processes and procedures are put in place to manage service delivery. The continuing growth in the use of pervasive devices will only exacerbate the situation. End-to-end monitoring systems are required, but people hold the key to service delivery.