Diagonal Scaling example: image processingĤ cores 8 cores (about the same CPU “usage” per box). Diagonal Scaling vertically scaling yourĪlready horizontal nodes Image processing machines Replace Dell PE860s with HP DL140G3s.Est limit/ ceiling limit current % days type # box units (total) (peak) peak left busy 120 db wait squid 18 950 req/sec 17,100 11,400 66.67% 48 (yes, fictional numbers) Fityk version: 0.8.2 output 2> guess Quadratic New function %_1 was created. Forecasting Automation jallspaw:~]$cfityk.You when this is actual flickr storage consumption from early 2005, in GB (ceiling is fictional) Is boring All we want is “days remaining”, so all we need is the curve-fit Use to automate the curve-fit Forecasting Automation Writing excel macros.When it is ceiling when is this? what you have left 15 servers busy apache proc limit = 1140 total procs ”Add a Trendline” with some decent correlation. Link to Chinese NewYear Photos (8% spike) (photo requests/second)įictional example: 15 webservers. Left “safe” ceiling CPU 85% total CPU = ~76 busy apache procs Like: database ceilings replication lag:ĭisk I/O wait for too much >40% creates slave lag* *for us,YMMV.To find ceilings Production: “it’s like a lab, but bigger!” “work” your resources will allow before degradation or failure Total CPU = ~1.12 * # busy apache procs (ymmv) Metrics App-level meets system-level here,.Per minute) (average processing time per photo) (apache requests) (concurrent busy apache procs) Metrics “Application” Level (photos processed.Store metrics in/out custom metrics easily compare lightweight-ish Iĭeployment and procurement easy and quick But clouds are still resources with costs and limits, just like your own stuff Black-boxes: you may need to pay even more attention than before Stuff Scalable Metric Collection/Display a lot of great deployment and management tricks come from them, adopted by web ops Performance for right now Measure what you have right NOW Don’t count on it getting any better Rules of Thumb Planning/Forecasting StupidĬapacity Tricks (with some Flickr statistics sprinkled in)īugs (disguised as capacity problems) edge cases (disguised as capacity problems) security incidents real capacity problems* * (should be the last thing you need to worry about).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |