Burstable Ram

Pommie · #1 (**permalink**) 16-07-11, 08:22 AM

I am having problems with my VPS because RAM usage is spiking and exceeding the 512 KB allowed. This is causing services to lock. I am told that the "solution" is to upgrade, but this self-evidently NOT a solution at all as it just raises the bar a little bit. I would still have services locking in an unpredictable way if RAM usage spikes. Just maybe less frequently. Your techies can't even tell me what is causing the RAM to spike.

In the short term I rebooted and freed up 150 K of ram so I will probably survive for a while before the problem hits me again. (This helped last time it happened about 6 months ago)

The reason I am posting is a plea to WHUK.

I really don't want to have to go through the hassle of moving hosts, and I like WHUK - your support isn't perfect but it's pretty damn good most of the time, but your policy of not providing burstable RAM is leaving me little option.

A quick google instantly throws up several competitively priced alternative VPS providers all offering burstable RAM. Now what would YOU do in my shoes?

I am running business on my web sites so I can't have random service blockages caused by Out of Memory problems - if you don't provide me with burstable RAM I will really have little option but to move to a supplier who does. And as I am pointing out they are not exactly hard to find! I wouldn't even object to paying (a little) more for the option if you offered it as an add-on

So is there any chance of a more enlightened approach or do I really have to leave you?

**John** · #2 (**permalink**) 16-07-11, 08:38 AM

We used to allow burstable RAM in the past but it created way too many problems. The situation being too many accounts were spending lots of time running over their guarantee and the nodes would play a game of deciding who could use what and when. It could be one day the node allows you your burstable RAM and the next day it doesn't as it has given it to someone else. The end result would be the exact same situation you have described, services being killed off.

As your problem is happening every 6 months or so and your usage is fairly high even after a reboot, I wouldn't suspect an underline problem with your scripts or VPS. It really will just be genuine usage.

In the next few months we will be launching a "Pay as you go" VPS concept. This will allow you to pick a base spec and bill you for additional time (every 5 minutes) you spend over that spec. This feature only works for Linux VPS and it involves auto scaling when resource demands are higher/lower.

We are currently still selecting the hardware this new platform will be running on, but within 2 months is a realistic time.

Pommie · #3 (**permalink**) 16-07-11, 08:47 AM

Hi John

Thanks for the speedy response.

Your pay as you go concept sounds like it might be a solution for me. Will it in effect provide burstable ram on a PAYG basis then? That seems to be what you are telling me but I just want to make sure

**John** · #4 (**permalink**) 16-07-11, 08:51 AM

Quote:

Originally Posted by Pommie

Hi John

Thanks for the speedy response.

Your pay as you go concept sounds like it might be a solution for me. Will it in effect provide burstable ram on a PAYG basis then? That seems to be what you are telling me but I just want to make sure

Hello,

You would basically pick a base spec, say 800 Mhz, 512 RAM and it would then scale up if your VM starts requiring more resources. So, in effect you would just pay for what additional you use. You will also be able to set a limit it can't go over as well, to avoid unwanted surprised at the end of the month

Oh, did I mention it will also be in a Cloud Environment?