Archive for the 'General Bloggery' Category

Merry Christmas!

We’d like to take this opportunity to thank all of our customers for their support and business over the previous 12 months, and wish everyone a very merry Christmas and a happy and prosperous New Year.

Instead of sending out Christmas cards this year we have again made a donation to Great Ormond Street Hospital for Children where over 150,000 sick children are treated every year.

We’d also like to particularly thank our NOC Technicians and Security Officers who are staffing both our LHC and Centro sites on a 24/7 basis over the entire holiday period in order to provide our customers with the continuous support they’re used to!

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Centro generator snow start test

As part of our extensive Planned Preventative Maintenance (PPM) schedule at our Centro datacentre facility, we run start tests on our generators every two weeks.

These generators provide stand-by power in the unlikely event of a utility outage on both HV supplies to the datacentre. We store enough diesel on-site to run the entire facility for around 4 days, and we have fuel resupply agreements with multiple fuel providers to ensure we can continue operating from generator power even in case of a prolonged outage.

We posted an image of the snow that had settled at Centro over on our Twitter page (over a foot deep in most places!), but as a generator start test was scheduled for today, we’ve uploaded the CCTV footage as an extra bonus:

Our generators are constantly heated and their starter batteries kept fully charged to ensure they are available as quickly as possible to provide backup power to our facility.

During this start test, the outside temperature was around -1C, and both generators started, synced and closed their breakers on to our main switchgear in 18 seconds, well within normal operating tolerances (we have battery supplies for our UPS system to last around 10 minutes before the generators kick in). The generators are run for around 30 minutes, during which time the on-board management computers are checked to ensure that there are no faults.

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Submarine fibre cable network maps

Yesterday, the BBC had a detailed article on the front page of their news website about a new high speed fibre cable being installed in to East Africa. Along with the article, there were some interesting network maps, provided by Telegeography.

We have a large Telegeography world submarine cable map on the wall in the reception area of our Centro datacentre facility, which always attracts a lot of attention and scrutiny from the slightly geekier of our visitors. The maps are also available to download as desktop wallpaper.

Centro reception area, with large network map

Centro reception area, with large Telegeography network map

On the subject of submarine cable systems, there is a detailed and fascinating (and quite long!) article by the well known author Neal Stephenson available  in the Wired magazine archives here. It’s certainly worth checking out if the subject interests you.

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

We’re hiring — PHP/MySQL developer wanted!

We are looking to hire a developer to work here at Gyron improving and extending our internal systems and scripts. You must have wide experience with PHP and MySQL, and some knowledge and experience with SNMP, XML, and interfacing with other code via APIs would be a bonus.

The role will be on a full time contract basis, based out of our offices in Hemel Hempstead, however there is also the possibility to work from home for some of the time.

Please send an email to jobs@gyron.net to find out more about the role. Don’t forget to include a bit about yourself and give us some examples of the work you done previously. Absolutely no agencies please!

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Centro Cooling efficiency

Early on in the design stage for our new Centro datacentre, we recognised the importance of in depth monitoring, reporting and trending capabilities across the entire facility, both for troubleshooting and early warning purposes, and for analysing efficiency data.

As such, the facility is equipped with an extensive Building Management System (BMS), which tracks around 2,000 variables in real time, including the status of the fire protection equipment (zoned detectors, VESDA, and FM200), leak detection system, cooling plant (CRACs, chillers, pumps/inverters), and power infrastructure (branch circuit metering, as well as meters on every outgoing way on main and distribution electrical switchgear, UPS plant and generators).

This level of monitoring has recently had even greater importance attached to it given the upcoming requirements for cap and trade carbon taxes, allowing us to collect data on energy usage (kWh, kW, power factor, voltage, and current) on a per-rack basis, and report this to our customers.

Our Centro datacentre facility has been up and running for nearly 3 months now, so we thought it would be worthwhile doing some analysis on the efficiency of our cooling equipment. We have installed highly efficient Computer Room Air Conditioning (CRAC) units, with electrically commutated (EC) fans, and have also used high efficiency over-sized inverter driven pumps, and high efficiency external packaged chillers with a higher than usual return water temperature, so we were hoping for promising figures.

Centro cooling efficiency graph

We’re pleased to announce that our initial analysis shows a PuE figure of 1.63! We are especially happy with this as our current situation with a large number of customers still populating their racks presents several big issues with part-load efficiencies on key plant equipment, particularly UPSs and Chillers.

The Green Grid specify:

Some preliminary work indicates that many datacenters may have a PUE of 3.0 or greater, but with proper design a PUE value of 1.6 should be achievable

We are continuing to improve the systems and processes in place at the Centro facility and hope to make even greater efficiency savings with additional investment in plant in the coming months.

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Snow!

It’s not so great news for those stuck on the roads, but on the upside, there are some great views from our datacentre in London today with all the snow settling on the ground!

Snow near the LHC datacentre

Snow near the LHC datacentre

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Network infrastructure upgrade sneak peek

Here’s a sneak peek to whet your appetite for our network upgrade!

Juniper MX240 routers, and EX switches in testing lab set-up
[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Economist article on datacenters and cloud computing

Just a quick post with a link to this article in The Economist about datacenters. It touches on the latest high density compute trends (including containerised facilities), power usage, virtualisation, and cloud computing services. Well worth a read!

We’re seeing more and more customers using our datacenter services as the foundation to build quite complex distributed computing and storage platforms on for use with virtualisation and so on — and the obvious next step is to layer cloud services above this to provide additional short-term capacity when required.

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

New deliveries at the Gyron office

We’ve had a number of exciting deliveries at the Gyron office this week.

First of all, the previously announced new Juniper network equipment arrived:

Juniper delivery

Juniper delivery

Shockwatch label

Shockwatch label

Tiltwatch label

Tiltwatch label

The equipment has been unboxed and checked, and set-up on our test bench to start testing before we announce maintenance to integrate it in to our core network infrastructure:

Juniper MX240 router with 40x1GE SFP interfaces, and a EX switch

Juniper MX240 router with 40x1GE SFP interfaces, and a EX switch

We also received a rather smaller package from Nike (this time without tilt/shock watch labels!). Unfortunately it was hard to work out which delivery our technicians were more excited about!

Gyroshoe

Gyroshoe

[Digg] [Facebook] [Google] [StumbleUpon] [Email]

Per kWh pricing for colocation services

There’s an interesting post over at the Green Datacenter Blog which references an article on computing.co.uk.

In it, the CIO for DTZ Holdings, a large real estate management company, recommends that customers stay away from per kWh pricing for colocation services for the moment, stating that due to the volatile energy markets at the moment customers may be exposed to massively increased power bills which they can’t forecast and budget for.

He’s heading in the right direction, but misses the central point. Colocation providers should exist to provide a colocation service to customers – their job is not to hedge their customer’s expenditure against price increases on energy markets. If colocation suppliers assume that extra risk then they’re actually hurting their customers – if the customer has no visibility or exposure towards the energy use of the equipment they’re using then there is no incentive for them to use more energy efficient kit, and colocation companies will just be charging more to cover their bases anyway.

Power Distribution Unit

Power Distribution Unit

In any case, customers are proactively coming to us and asking for pricing on a per kWh basis. It’s the only sensible way to compare pricing from different facilities which operate at different power densities. Customers don’t look at their requirements and say “well, it’s obvious that we need 500sq ft”. They look at their requirements and say “well, that’s 60kW of IT load, what do we need in order to support that?”.

We’re by no means the biggest colocation provider in the world, but even at an average load of 1MW at the moment, we still eat through around 730,000 kWh of electricity per month. Anything we can do to lower that figure is both good for us on a financial basis, and is good for the environment. As such, all of our large colocation deals specify a rental figure for the floor space, which tends to be fixed over a period of time (with RPI+X% capped increases once per annum), and a separate figure for kWh of power used, which is backed off against the utility power charges. We make our money from the colocation service itself, not from marking up power charges!

[Digg] [Facebook] [Google] [StumbleUpon] [Email]