Have your say at computerweekly.com
Companies can plan for London network disaster
The test performed in the City of London to assess the vulnerability of communications links (Computer Weekly, 30 November) highlights the importance of ensuring network resilience and availability in the event of a disaster. This has an effect on the rest of the UK as London serves the heart of the UK infrastructure. Should London be wiped out, the UK would lose 90% of its infrastructure.
The comment made by Neil Robinson that "risks are outside companies' control" is misleading as there are numerous ways for companies to back-up critical communications channels to safeguard against unplanned downtime.
Suppliers do run networks through the same infrastructure, which makes it all the more important for companies to buy diversified services from different providers.
Use more than one ISP for bandwidth. Many companies may think they are using more than one provider, but ultimately they rely on the same one because telcos buy bandwidth from the same small number of carriers.
Although companies should expect their communications providers to be more rigorous in their approach to customer and network service, there is also an onus on organisations to assume some responsibility for network resilience and availability issues.
Keith Tilley, UK managing director, SunGard Availability Services
MPLS is not a one-stop network solution
Yankee Group's prediction that the use of multi-protocol label switching (MPLS) is set to double in four years (Computer Weekly, 7 December) indicates that network performance is a priority for many businesses. It is positive to see firms adopting these measures to bolster the performance of their business-critical applications.
This is, of course, good news for carriers selling MPLS, as they will ultimately increase their profits by charging the customer more for using MPLS with a classic "bait and switch" selling technique. Although every carrier offers different tariffs, it is not uncommon to find that the initial MPLS class of service offers savings of 10% to 15%, but that adding further classes later in the contract can increase costs by 25%.
In addition, MPLS in isolation does not address the entire application performance challenge. MPLS does nothing to fix the problem of too many applications contending for limited bandwidth between the business and the carrier network.
Only by having a holistic view of the network and monitoring applications will products such as MPLS be able to fully deliver the benefits intended at the price expected.
Roger Hockaday, director, Packeteer
Kiss and give clients software that works
In response to the article "Avoid adding risk and complexity to projects to achieve goals rapidly" (Computer Weekly, 7 December), I fully endorse the views expressed by John Wailing with regard to keeping software projects simple. The example of one of our project rescue cases is enlightening.
In this particular case we re-implemented the software required for a project that was going seriously wrong with the usual symptoms of being late and not working. The time taken to re-implement was some 15 times less than it had taken to get the original software to the stage where it was cancelled. The reasons for this success map neatly onto the reasons cited in the article.
- Keep things simple. This is the Kiss adage from real engineering saying "keep it simple, stupid". Our team only provided what was really needed, it did not waste time or brain power creating functionality that was unlikely to be used.
- Do not reinvent; try to reuse. Although this was a re-implementation, the reusable components we had meant that a large portion of the software was already in place before we started.
- Do not spend too long on requirements. Knocking up a quick prototype to show immediate progress was a boost for the client's flagging morale.
- Finally, surprise people. The thing that most surprises people is when software actually works.
Dave Knight, managing director, Igence
Why don't websites have contact numbers?
Has Computer Weekly or its readers noticed how rarely one can find a useful contact telephone number on websites?
Many websites have no contact phone numbers anywhere, or only provide unhelpful numbers for automated response services which leave you stranded if your query does not fit into the algorithm to which they work.
Is there some plot afoot to bring about the demise of the telephone? I think we should be told.