Peak and off peak hosting hours opinions

  • RealMetrics
  • Newbie
  • Newbie
  • RealMetrics
  • Posts: 7

Post 3+ Months Ago

We monitor hosts for uptime and webspeed and we going to split our uptime metric into a Uptime During Peak and Uptime During Offpeak to avoid penalizing hosts that have scheduled maintenance during Offpeak hours. I'd appreciate some feedback on the hours of the day that each of you would consider "peak". For example, you could say that from 9:00a-6:00p CST is peak and every other hour is less critical. Or, you could say that hosts should only do maintenance from 1:00a-4:00a CST and everything else should be considered peak.

Opinions?
  • Daemonguy
  • Moderator
  • Web Master
  • User avatar
  • Posts: 2700
  • Loc: Somewhere outside the box in Sarasota, FL.

Post 3+ Months Ago

That's arbitrary. Depends who your customers are; do they do business in other geographies? AP, EMEA? If so, there's no such thing as 'non-peak' time, technically speaking. There may be a period of time that historically sees *less* traffic than at other times, but if a company is global that's not really a good metric nomenclature to use. ;)
  • RealMetrics
  • Newbie
  • Newbie
  • RealMetrics
  • Posts: 7

Post 3+ Months Ago

Good point! Let's say we go with "Scheduled Maintenance". What are the hours that seem more acceptable that others?

What we're looking for a way to differentiate between a host who schedules maintenance during lower volume times and a host who just goes down for no reason at all. Should we display an hours downtime as equal regardless?
  • Daemonguy
  • Moderator
  • Web Master
  • User avatar
  • Posts: 2700
  • Loc: Somewhere outside the box in Sarasota, FL.

Post 3+ Months Ago

RealMetrics wrote:
Good point! Let's say we go with "Scheduled Maintenance". What are the hours that seem more acceptable that others?

What we're looking for a way to differentiate between a host who schedules maintenance during lower volume times and a host who just goes down for no reason at all. Should we display an hours downtime as equal regardless?


That's a sticky subject, especially to me.

In my opinion, even scheduled maintenance downtime, IS downtime. All too many hosts do not include this in their uptime guarantee. If they mention it at all, it's a caveat; "99.9% -- apart from standard maintenance". I disagree.

When *I* tell a customer "nothing less than 99.99999%" I mean that. There are NO maintenance windows, no scheduled downtime ... it's just there, all the time.
  • RealMetrics
  • Newbie
  • Newbie
  • RealMetrics
  • Posts: 7

Post 3+ Months Ago

Thanks! I see your point about how peope can misuse the term "scheduled maintenance" to their advantage.
  • RealMetrics
  • Newbie
  • Newbie
  • RealMetrics
  • Posts: 7

Post 3+ Months Ago

Thanks for the feedback! The International angle is important to us also, so your point is well taken. That seems to be the biggest barrier to implementing the two different metrics.

We originally were thinking of using two separate graphs, one for peak and one for off-peak. But deciding what is off-peak is the challenge.

Our latest thinking is to give the user the option to pick their off-peak times from a drop down menu and compare hosts accordingly. How does that sound?
  • Daemonguy
  • Moderator
  • Web Master
  • User avatar
  • Posts: 2700
  • Loc: Somewhere outside the box in Sarasota, FL.

Post 3+ Months Ago

If you allow them to choose a window of opportunity, you have to ensure that no other customers with dissimilar windows are shared on the infrastructure.
That is not an easy thing to do when you are a relatively small hosting company. You can't have eight sets of hosting gear on the raised floor, one for each three hour window; it's not financially practical. ;)


Problem number 2; What happens if the maintenance does not get completed in that window? Which means you need a much larger window, which in-turn should affect your uptime guarantee. It's a vicious cycle.

You could use the industry standard disaster recovery model; 2 sites in different locations using different backbone providers. Exact same gear, exact same applications. One is in hot-standby. Do regular data shifts, and when you require maintenance, you shift traffic (using whatever mechanism you prefer, such as hard DNS, VIP controlling, Network Dispatching, routing, etc. et al.) and do your maintenance with NO windows.

When you are done, you move it back. That's a costly venture though; purchasing two entire infrastructures and the bandwidth. Ouch. :) Of course, you could have the second set on the same raised floor and forego redundant BW.

Tough call.

Post Information

  • Total Posts in this topic: 7 posts
  • Users browsing this forum: No registered users and 9 guests
  • You cannot post new topics in this forum
  • You cannot reply to topics in this forum
  • You cannot edit your posts in this forum
  • You cannot delete your posts in this forum
  • You cannot post attachments in this forum
 
cron
 

© 1998-2014. Ozzu® is a registered trademark of Unmelted, LLC.