Oaksys Tech Update

I’ve been working in the computer industry for over forty years. It is not often that I see an implementation of technology where I think “Wow!” At a recent exhibition I came across GID-Quantor’s Silent Cube technology. These devices use WORM (Write Once Read Many) optical disks to store vast amounts of data. You can store Peta byte (1000 Gigabytes) of data, on-line, in a three rack configuration.

The devices have been carefully engineered with a high level of redundant resilience and active error monitoring. They will also support remote site mirroring as part of the package. They are also engineered to be very low power usage, so the running cost is low; just two Watts for 8 Tera bytes. They are also highly modular.

If you have high volumes are static archival data that needs to be retained long term, but with rapid accessibility requirements, this may be the technology for you.  GID-Quantor have a history of providing archival microfiche solutions, their experience shows in the design of these Silent Cube’s. I can see many uses for this technology in the financial markets; for example, voice recording, archive of price feeds/trades for compliance purposes.

Alaric

The attack on Google Mail for some Chinese dissidents was based on a security weakness in Microsoft Internet Explorer version 6.  It demonstrates the need for keeping your software up to date. One (free) tool that is highly useful in achieving this can be found at Secunia. It examines your software version level, of most common software, and advises you if the current version is out of date. Usually it offers a location where updates can be found. As a consequence you are likely to maintain the essential security patches.

Anti Virus and anti-malware  software can only go so far in protecting your machine. If your PC accesses the Internet or receives email you need to make sure that all of your software is kept up to date with the secuirty patches.

I attended a seminar where Microsoft’s latest tools for managing Hyper-V were demonstrated. Virtualisation is creating an increase in the complexity of the Server environment. Businesses will need increasingly skilled personnel to manage the environment. They will need good software tools to be able to perform effective management of the environment.

The other thought which struck me is the need for operations management to have good tools to measure/track the operational cost (and cost recovery) of running applications in a multi-server data centre. It takes considerable capital expenditure to set up a full loaded Microsoft/VMWare environment. The servers have to be powerful as does the storage environment. Recovery of the capital cost during the financial depreciation period will be an additional loading on the operational costs. Fortunately such software tools do exist to help track the costs and to bill the user business. More on this later.

Guru

In a recent expansion of Telecity’s datacentres to Powergate in North London they used Geo to provide fibre through the sewers. Given that one of COLT’s major hubs is located in Powergate one wonders if Telecity made use of COLT fibre as well as Geo.

Guru

My 25 year old daughter has flown the nest of the family home. She has her own career and her own home some 15 miles away from Lewisham in South East London. When vandals struck at the local power supply sub-station they took out the power to 60,000 people. My daughter was without power in her home for 4 solid days. She was able to come and stay back in her old bedroom, so the power outage did not affect her too much. She had a standby arrangement.

I wondered how well the local businesses fared with the power outage? Network connectivity dubious, no power for servers for 4 days, even charging mobile phones was a problem if you were stuck in the  the location of the outage.

I wonder did any of the businesses what they would do in the event of a regional power outage. Rehearsal and problem analysis is an essential part of the BCP process for those businesses. It is bad for customer relations if you don’t have effective BCP, but conversely your company will stand out if it continues its operations in the face of a regional outage.

Virtualisation of Servers is not new. This concept was in use on mainframe computers back in the 1980’s. The re-emergence of virtualisation to optimise the number of servers in the equipment racks has highlighted an old problem. As more users/transactions share a single computer the amount of disk I/O can become a bottleneck.

The success of Virtualisation replacing standalone servers depends on  the low CPU utilisation by most application processes. However virtualisation and the higher capacity of modern disks also tends to reduce the number of hard disk spindles available to share the I/O load of each user.

If a 1000 users used to share 10 distributed servers and this load is transferred to one multicore/multiprocessor server they are likely to start encountering delays due to disk I/O.  A lot can be done with solid state disks and disk virtualisation, but it requires thought and metrics.

Guru

The latest Web 2.0 and SOA seems all about getting things done quickly rather than producing robust computer systems which are secure and producing auditable results.
A new entrant to the IT world that I know is being given projects connected with financial systems. They give him projects in a range of different programming languages even if he does not know that language. He is then expected to pickup the programming language in a week or two and to generate public facing production systems from that language within a few weeks. So far there have been no tears, but the potential for corporate embarassment is substantial.The person(s) at fault is this young man’s boss(es). I think the time has come for legislation that any public facing system involving credit cards to be certified by a qualified chartered professional before release to the public.

Guru