FlexiScale Blog

One of our customers who has had a lot of constructive feedback for us, and has spent a lot of time playing around with FlexiScale functionality and the API has got a great blog post up at http://www.matchpeg.com/blog.asp  they have also released the various tools they have built along the way, including their own web based control system for FlexiScale, using our API.  These are available at http://www.matchpeg.com/brochure/misc/flexiscale.asp

I think the work they have done is brilliant, and we’re certainly looking at incorporating some of their ideas in our redesign of our web interface (which we know is inferior compared to the rest of the service) which will happen soon.

Tony.

This is a first for us at FlexiScale. Tony manages to get quoted in the Register during his participation in the Cloud discussion panel at Structure 08. There’ll be no dealing with that ego now

We’re all excited at XCalibre as FlexiScale has been named as a finalist in the prestigious 2008 TechWorld Awards. The category we’re competing in is ‘Virtualisation Product of the Year’ and FlexiScale is up against some big names including VMWare and NEC, but that just convinces us of what a terrific product we’ve created. The winners will be announced on 11 June 2008, so we’ve got the champagne on order just in case.

Tony

Well there’s been a lot happening in Camp FlexiScale since Tony’s last post.

The developers have worked hard to bring more choice in OS image and we now have the following:

Windows Server (32 and 64-bit)

CentOS 5 (32 and 64-bit)

Debian (32 and 64-bit)

Ubuntu 6.06LTS (8.04LTS to follow soon)

Most of these images have been available because you guys have asked for them. We’ve always been and always plan to be customer led when it comes to development. We feel it’s important that you guys have your say which is one of the reasons that Jonathan has created the “wish list” thread on the FlexiScale forum. Of course, we won’t be able to cater for every request but where there’s clear demand we’ll certainly do our best to give you guys what you want so by all means go on and have your say.

FlexiScale is now growing at the phenomenal rate of 1% a DAY!

So how come that we’re getting this growth yet my phone’s gone quiet?

Online signups!

Since the online signups went live around 90% of people are signing up themselves online. This is great as it really shows how easy and intuitive the whole process is from signing up to provisioning and utilising your servers. Exciting times!

Please remember though that there are still some real live human beings at XCalibre that are happy to talk to you and help you with anything you need or any questions that you’d like answered. Besides, you’ve usually got some really interesting projects on that we miss hearing about.

We still want to make sure that our new online signups are getting on ok so what we’ll be doing is calling you. You don’t need to panic and hide from your phone, all we’ll be after is a friendly chat to make sure that you’re getting on alright and have everything that you need. So we hope you won’t mind.

We’ll be updating the blog on a more regular basis to keep everyone posted on where we are and what we’re up to so watch this space.

John

The last couple of weeks have certainly been a bit of a rollercoaster ride here, although it’s finally ending on a very positive note.

For several weeks now we’ve had a significant problem when we were restarting physical servers, in terms of how long it was taking them to come back online, due to the initialisation they needed to do. This was managable when we only needed to restart one server, but if we needed to restart multiple servers it became rather frustrating.

This has compounded the other problems we have had in the last couple of weeks, where on several occasions (for an upgrade, a power outage and then a switch replacement) the entire platform needed re-initialised (either bit by bit, which we can do without service interruption, or completely). Until yesterday this process could unfortunately take up to 7 hours or more to happen. I’m very pleased to thus let you know that this problem has now been completely fixed due to some innovative and rather clever work by our engineers, and the initialisation of a server now takes 30 seconds.
This should ensure if we do have any problems in the future (fingers crossed, but sod’s law is fairly hard to avoid!), that we can recover from them very quickly.

So, on to yesterday’s problem.

There was a very brief (a few seconds) power outage at the main datacentre we use for FlexiScale, caused by human error, which we have been reassured won’t happen again as the process that was happening is being modified to prevent this.

This caused a spike to hit some of our equipment, and although the vast majority (some 100 servers) all came back ok, we started to see some intermittent issues with our core FlexiScale switches.

I should point out at this time that the switches were in a redundant configuration, and we did have an arrangement to obtain additional switches should one fail within a matter of hours. We didn’t consider both failing at the same time a realistic risk, now we know better.

The switches were still functioning to a degree so we left them running whilst we got the two replacement switches delivered. (Which involved yours truly being the courier for them to speed up the process!). These then needed installed, configured and then patched into the network which duly happened, and then the platform was brought back online.

Needless to say we have learnt a lot from this last few days, here are a few of the things we have achieved or are going to be changing:

• We’ve upgrade the software running the system to a newer version, which has a lot of improvements in the stability of individual servers.

• We’ve fixed the problem with initialising servers, which will help enourmously in the long run.

• We will be investigating powering parts of our cage from different sides of the datacentre to ensure maximum redundancy (including the switches being on completely seperate feeds!)

• We will be working out a better plan for coverage of key equipment (even in cases where it is in a redundant configuration) to ensure multiple failure situation’s can be dealt with more effectively.

Overall I’d like to say thankyou for the support we’ve recieved from customers during this time, and we look forward to continue bringing you more innovative features, and a highly reliable service in the future. We have some very exciting features being released over the next few months, and look forward to showing you them.

Tony Lucas

Chief Executive Officer

Back in 2004, January 11th to be exact, a website we hosted on one of our shared webservers got mentioned on Slashdot, and promptly caused the webserver in question to be completely overloaded, as 10’s of 1000’s of people tried to access it (to make matters worse, it was very image heavy as well!). Read the rest of this entry »