Welcome to the High Availability Hosting Blog

Yesterday (22/08/2010) we successfully tested and failed-over the Sheffield Data Centre to backup generator power.

The culmination of over a years planning, and many hundreds of man-hours to implement a N+1 power system without downtime or service interruption was finally completed just after 1pm yesterday. Utility power was disconnected, the Automatic Transfer Switchgear started the generator, waited until it had stabilised, and then automatically threw the switches transferring the feed.

Testing over the last three weeks, including a battery of fresh testing yesterday ahead of the fail-over included auto start/stop, continuity, voltage & frequency, load testing, UPS testing, engine servicing etc…

The failover was repeated twice and we had the building totally generator powered for 15 minutes before switching back to utility (mains) power.  The system is now set to automatic so in the (unlikely) event of a power cut we’ll fail-over to generator without human intervention.

Regular testing will ensure we’re always ready!

We did video the failover and take some stills, they’re in the process of being editited and we’ll put some images up on the blog soon.

The Sheffield Data Centre is now an N+1 power facility.

We finished monitoring the Low Latency Network (and the colocation networks as a whole) for bandwidth usage last weekend.

We’ve now analysed the reports and are forming packages which better suit the usage by customers, and the data centre as a business.

Customers will be contacted this week with an update. In the meantime the bandwidth analysis will remain so we can begin trending the network and available capacity.

Just a quick update to say all the planned weekend works to cable and connect power to the generator charging system were successful with no service outages.

Ongoing works this week include installing a power usage meter and the emergncy stop breaker to the Automatic Transfer Switchgear. All non-service affecting.

Once the electrical works are complete we’ll post some pics!

Cabling works for our standby generator are finally underway, and should be completed over the weekend.

We have plans to connect the Automatic Transfer Switchgear to the generator TODAY (06/08/2010), and connect & test the power to the auto-start system over the weekend.

Next week we’ll run a series of tests of the auto-start system, transfer controllers, and the generator iteself (without switching supplies), with a view to a full failover test toward the end of August.

Later than originally planned – delays have been caused by human resources (lack of people time!!), weather, ongoing client projects, and other data centre engineering works.  We’d like to thank everyone who’s showed us their support and kept a keen interest in our expansion!

All client and core equipment is protected by UPS anyway, but the generator will add an extra layer of redundancy should there be an extended power outage.

Regards,
Stuart.

www.sheffielddatacentre.com

Over the weekend we’ve updated the 3 x High Availability Hosting Limited websites:

www.hahosting.com
www.thevirtualcloud.com
www.sheffielddatacentre.com

The main updates are over on TheVirtualCloud – where the full Flexible Cloud Servers information is now live, along with prices, examples etc…

Take a look at http://www.thevirtualcloud.com/cloudservers/default.aspx for all the details!

Other updates include links between the sites, a full list of all services (all sites) on the homepage of www.hahosting.com, and the removal of legacy VMware hosting.

We’ve started to gather statistics on bandwidth usage on the Low Latency Network used for gaming.

We currently provide an “all you can eat” service with no QOS or Traffic Management for £5 per month.  Unfortunately due to massive demand and unfortunately some abuse we’ve had to re-think this strategy.

Later this year we’ll be introducing an allowance on the connection, and charging (at a low rate) for additional usage thereafter. We’re using MRTG to monitor port usage between now and August, and will use this data to give us an average baseline for and come up with a generous inclusive allowance for our gaming customers.

MRTG is non-intrusive and won’t impact your gaming performance.  Stats will be available to each customer, along with throughput graphs and totals.

The “Low Latency Network” will remain just that – a network which is logically closer to the WAN switches and devoid of Traffic Management.  Unfortunately the 100% free bandwidth could never last forever, and was actually discontinued for business customers earlier this year.  We hope establishing a baseline of actual usage rather than plucking a number out of thin air will be a fair way of determining the allowance.

I’d like to also remind everyone that Web Hosting on the Low Latency Network is strictly prohibited.  Gaming control panels are fine, but general hosting & downloads is not allowed as this eats into the bandwidth allocated just for gaming.  Any user found web hosting on this network (we do check) will be instantly cut off and asked to discontinue.

If you’ve got any questions you can email support @ hahosting.com where we’ll be happy to help!

17/06/2010 04:30 – SharePoint is now back online.  Due to issues moving the master config databases to a different SQL installation, plus issues installing a SharePoint rollup patch we have decided to roll-back all changes made by un-doing the VMware snapshot.  No data was damaged and SharePoint is back as it was before the change started.  We’ll review the data we collected and re-attempt the upgrade soon.

17/06/2010 00:30 – SharePoint hosting is now currently offline while planned upgrade works have started.  We aim to have the service back and running within 4 hours.

The scheduled work to upgrade our Client Control Panel this weekend was completed successfully.  We’ve upgraded from WHMCS 3.7.2 to 4.2.1, and made improvements to the overnight Scheduled Tasks, the invoice and email layouts, as well as integrating the control panel into the “look and feel” of the rest of the HA websites.

The Client Control Panel is now accessible via the new link of https://clientaccess.hahosting.com/ccp/

The previous links (ccp.hahosting.com & secure.hahosting.com/ha/ccp/) will still work but redirect you to the new link.

No more “visible” work on the Client Control Panel will happen now for a few weeks while we gather feedback and fully use/abuse/featuretest the new platform.

The next job on the clientaccess platform is the go-live of the Hyper-V Web Access control panel, powered by WebsitePanel, currently undergoing final testing.  Watch this space!

Just a reminder that the Client Control Panel (support tickets, invoices, client management, product management etc…) is being upgraded this weekend, (29th & 30th) and will be unavailable for intermittent periods while we perform various upgrade works.

Clients were notified last week, but just in case you missed it and was wondering why WHMCS (the Client Control Panel) isn’t working, this is why!  We expect works to be completed over today & tomorrow, with full service restored for Bank Holiday Monday.

Support tickets will still be raised (although there may be a delay) as normal by emailing support @ hahosting .com, emergency support is always available on-call via 0114 2280022.

When the work is completed, the Client Control Panel will be available at the new URL of: https://clientaccess.hahosting.com/ccp/
Although the old URL’s will continue to redirect for the next few months.

Over the coming months, we’ll be migrating all our control panel services onto the clientaccess.hahosting.com domain, centralising admin tasks to one area and avoiding confusion from multiple domains.  We will keep users posted as we move forward…

We’ll post another update when the work is completed.  Follow the RSS feed for instant updates direct to your RSS reader!