SLAs offered

[kogin]

Your service is very interesting. Can you share more about the SLAs you offer?

Amazon EC2 does not seem to have any SLAs around their service. What do you offer around availabilility of servers, network, capacity, response times etc?

[lazzurs]

Hi,

We will be offering SLA's on the FlexiScale platform however the exact figures are still being worked on internally.

As soon as we have defined these SLA's internally I will reply to this post with the figures however what I can say now is we have designed the platform for '5 nines' (99.999% uptime).

Finally I can also say the SLA we offer will be higher than the 99.9% we currently offer to our dedicated server customers ;)

If there are any further queries we can offer please let us know :)

[bostonpaul]

I am seriously considering your service...any updates in regards to an SLA?

Thanks,

Paul

[tonylucas]

Paul,

Apologies for the delayed reply, somehow this post got missed.

We are hoping to publish the SLA by the end of January, it's a little more complicated to work out than a normal SLA, as you need to take into account recovery times etc.

If you do have any specific questions feel free to ask them here or drop us an e-mail.

Regards,

Tony.

[gegard]

Although high SLA is important, I would be interested in knowing what the cost implications of higher SLAs would be. Presumably 99.999% would cost more than 99.99%: if that's (say) 1% more than the current base price then we'd consider it, if 100% more then probably not.

It is important to know when hardware has started failing so that you can choose not to commit a failed transaction rather than accumulate bit errors that you don't know about. Disk errors are presumably already seen and managed this way but do all servers use ECC memory so the same can be done for RAM errors?

Thanks.

[tonylucas]

Hi,

There will only be one uptime SLA level, although multiple different levels depending on pricing is something we may consider in the future for customers.

I would imagine though if we were to do this the price difference for the improved SLA (above 99.99%) would be quite exponential, as the closer to 100% you need to achieve the far higher cost.

I can confirm all the servers are using ECC Ram.

Regards,

Tony.

[gegard]

... if we were to do this the price difference for the improved SLA (above 99.99%) would be quite exponential ...
That's what I thought: I'm not so concerned about the uptime, rather the ability to notice a failure, to prevent a failure from causing data corruption, and to recover from the failure.

... I can confirm all the servers are using ECC Ram.
Excellent - that makes sense!

Regards.