Quick update on Rackspace costings


#1

I just had a quick look at the Rackspace account, and it’s quite interesting. TDLR is that it’s going the right way :stuck_out_tongue:

Our billing period is to the 7th of the month, so right now I’m working on the 7th Nov -> 7th Dec data. I’ll have the Jan bill in a few days, of course. I haven’t yet gotten to the point of writing shiny R webapps yet, but my scripting is improving, so I wrote something quick that takes in the Rackspace CSV and spits out a summary:

> rackspace('/tmp/mozilla_greg0/020-864112_B1-29142514.csv')
                     Type Charges
5    NG Server IP - Hours       0
3            Server Image    2.10
2              CBS Volume  107.40
4 NG Server Bandwidth Out  556.51
6        NG Server Uptime 1155.23
1                   Total 1821.24

So, noticeably, we’re under our $2k budget for the first time in ages, and with the CDN running now, I expect that to drop further (good stuff @evgeni :stuck_out_tongue:). That’s about $60 less on servers (within monthly fluctuations, to be honest), but $150 down on bandwidth.

Given the CDN wasn’t activated until Dec 12th, I’m struggling to explain it, but hopefully it’ll stay low now that the CDN is in. More data in a few days, and then maybe we can add another slave or two (I’m calling shotgun on the first one for metrics stuff :slight_smile:)


#2

The CDN isn’t that active yet - only on downloads and stagingdeb which aren’t very high in traffic.

One possible explanation could be that a lot of us developers took it easy over Christmas leading to less testing but that’s pure speculation.


#3

Yeah I thought about that, but I’d expect that in the coming bill - this one is only to 7th Dec. Maybe Thanksgiving has an effect too…


#4

Here’s this month’s data (7th Dec to 7th Jan)

                Type Charges
NG Server IP - Hours       0
        Server Image    2.17
          CBS Volume  110.98
Server Bandwidth Out  493.88
    NG Server Uptime 1194.62
               Total 1801.65

Definitely looking good. I’ll work out what we’re spending per slave when I have a moment and see if we can spin up any more yet.


#5

I’m looking at our systests and it does appear we run additional tests now since we created the new nightly promotion pipeline (which runs tests in a rackspace VM) but didn’t remove the old systests. My patches are almost done to unify it all on forklift. Hopefully this week I can summarize the situation and the matrix we do want to test.