London Datacentre Relocation
Wednesday, November 15, 2006
London Datacentre
Planned Relocation to a new Data Centre
We plan to move all of our setup to a new London data centre this month. This move affects all customers, and this update contains essential info about the move.
All of our kit will be moved out of Telecity / Redbus Sovereign House. The new data centre is less than a mile away at Greenwich.
This move is due to a number of serious events that have taken place during the last year. These serious events are set to continue into the future with Redbus.
Why relocate to another data centre?
1. Telecity / Redbus merger.
With the merger of Telecity and Redbus, this has resulted in significant price increases (approximately 400%) and a reduction in service.
On-site engineer numbers have been scaled down, leading to long waits when remote hands are required. With Telecity continuing to expand, another round of price increases and staff reductions is expected. Telecity Redbus are also unwilling to commit to contract terms longer than a few months, which is not ideal for long-term business planning.
2. Space for growth.
Existing Redbus London Datacentre sites are almost full, and additional space is at a minimum. Any space that is available is priced extremely high since the merger. This makes it impossible for us to continue to grow without passing on price increases to our own clients. Up to now, we have soaked up the increase in our costs, but with this set to continue, we're unable to absorb the increases without taking action. To continue growing, we need to secure additional space. Avoiding 6 or 12-month waiting lists or having to spread our equipment between different buildings.
3. Electrical Power.
They have also introduced new power policies for the amount of power that can be configured per rack. This has resulted in us not being able to get anywhere near to the ideal capacity of a colocation rack. The latest limit per rack at Telecity Redbus is eight amps, which is enough to power approximately 10 servers. (This is in a full rack capable of physically housing 40 1U machines.) At the new facility, all racks will have 16 amps as the standard, with additional upgrade options available. This will allow us to continue getting the most from our colocation rackspace.
To summarise:
If we stay at Redbus data centre, we would be paying 3-4 times more per colocated rack. With 3 times fewer servers per rack and decreasing remote hands coverage. Combined with the low contract times, prices could increase and policies change for the worse (as we have seen several times within the last year) at any point.
At this point, we wish to stress that this decision is not solely based on our costs. We have always focused our services on high quality and have the same in mind when seeking our own upstream providers, but to continue at Redbus, it would be impossible to continue offering the same quality and the same prices, and we would have no option but to increase prices, which is something we are keen to avoid.
When is the move happening?
All London Datacentre equipment will be moved on the nights of the 17th and 18th November 2006. The work will be done in 'batches' whereby the equipment will be split and done in different stages. We will ensure one batch is fully functional at the new facility before starting on the next. The first batch of each night will start at 11 pm, and 2.5 hours is scheduled for each batch to be completed before the next batch is started, with the last batch of each night ending at 5 am.
Please note that 2.5 hours of downtime per batch is the absolute maximum limit. If all goes to plan, downtime shouldn't exceed half of this amount. At this point, we do not have specific schedules for each server, and this information will be provided closer to the time. All clients should subscribe to our helpdesk announcements to be kept informed of the specific times and batch information.
How will it be done?
A specialist server relocation company is being brought in to handle the moving of equipment. There will be a 13-man team of qualified technicians on site performing the move. They take care of power downs, packing (all equipment will be packed in bubble wrap), moving and everything else associated with the work.
This company has years of experience in work like this and has performed many successful moves for companies such as Barclays. We know there have been some high-profile data centre moves at other web hosting companies that have gone wrong. Those involved experienced a lot more downtime than planned, damaged equipment, and similar issues.
We would like to assure everyone that all measures have been taken to ensure that absolutely everything goes to plan with our move.A great deal of data centre migrations happen without any kind of problem but generally, we don't get to hear about these as they aren't as newsworthy :)
What is the new data centre like?
The physical infrastructure in the new data centre is better than the current facilities at Telecity Redbus. There will be absolutely zero drop-in facility standards (actually the opposite). In addition, the new facility has provisions for building solid brick walls around certain areas of the data floor for additional security (and heat protection) as opposed to Redbus, where metal cages are used instead. Along with the solid walls between suites, the new facility allows for dedicated air conditioning and fire detection/suppression equipment per block. This means that we will not be solely reliant on the data centre's facilities in this area.
Will servers be required to move again in future?
No. A significant aspect of this is that multi-year, fixed-price contracts are now an option, allowing for long-term planning and security. The situation at Redbus is such that contracts longer than 3 months are not possible, which allows for a whole series of short-notice price increases in the future.
Will the network change?
We are not making any changes to the network in any way, apart from the physical location of the equipment. The same bandwidth providers, equipment, IP addresses and everything else will continue to be used. The excellent network uptime we have experienced since day one will continue into the future.
Both the new facility and Redbus are carrier-neutral. This means that those colocating in the buildings handle their own network without this being provided by the building. This effectively means that a provider in one cage will have a completely different network setup (and often quality) than a different provider just a few metres away. Therefore the building a company has servers in (when dealing with carrier-neutral facilities) has absolutely no bearing on the quality of the network of that company. With our high-quality network providing high uptime so far, this will continue at the new facility with no change.
Will there be any changes in performance after the move?
The only performance changes are good ones, due to us being able to continue growth with the same high quality we have provided so far. Without this update and the downtime involved, nobody would be aware that this work has been done. Our servers remain unchanged, the network remains the same, and traceroutes are identical, etc.
We hope that the info in this notice has helped anyone worried by the words "data centre move"? We can assure everyone that this is a positive step forward in our continuing provision and growth of high-quality hosting services.
If you have any questions or concerns, feel free to contact us.
[UPDATED]
Our planned move was finished last night, with our last server coming back online just after 4am Sunday morning. Everything went according to plan, and we stayed within our maximum allocated downtime estimate on both nights.