Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Performance Tuning Amazon ec2 and RDS
Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Type||Name||Latest commit message||Commit time|
|Failed to load latest commit information.|
== ABSTRACT As organizations deploy more of their infrastructure into Amazon's Web Services, there is a growing need for guidance on scaling MySQL in the cloud. This presentation will address performance challenges and management issues in the cloud. By the end of the presentation, attendees will understand cloud-specific advantages, scaling challenges and how to achieve high availability on that environment. == TRACK Database Administration == EXPERIENCE LEVEL Intermediate == Slides ***** Availability Zones ***** Eight Global Locations US (x3) EU (Ireland) Asia Pacific (Singapore, Tokyo) South America (Sao Paulo) AWS GovCloud (US Gov't & Contractors) ***** Availability Zones ***** - Different Performance Characteristics - Different Hardware - Different Pricing ***** Availability Zones (Strategy) ***** Split services (hot-hot) between US-EAST & US-WEST NEED MORE ***** Instance Types ***** Micro DB Instance$0.025 Small DB Instance$0.105 Large DB Instance$0.415 Extra Large DB Instance$0.830 High-Memory DB Instance Class Extra Large DB Instance$0.585 Double Extra Large DB Instance$1.170 Quadruple Extra Large DB Instance$2.340 - $1600/mo ***** Instance Costs ***** ***** Costs of comparable Physical Servers ***** == Links == http://www.quora.com/Amazon-EBS/What-are-some-tips-for-tuning-EC2-EBS-IO-performance http://perfcap.blogspot.com/2011/03/understanding-and-using-amazon-ebs.html http://victortrac.com/EC2_Ephemeral_Disks_vs_EBS_Volumes http://stu.mp/2009/12/disk-io-and-throughput-benchmarks-on-amazons-ec2.html http://www.appdynamics.com/solutions-java-monitoring.php http://www.whenisfive.com/2012/06/09/performance-tuning-lamp-stack-for-ubuntu-12-04-lts-ec2-instance/ http://serverfault.com/questions/338262/tuning-javavm-on-aws-ec2 http://serverfault.com/questions/197224/ec2-regular-performance-issues-without-obvious-resource-contention http://harish11g.blogspot.com/2012/03/varnish-aws-ec2-performance-tuning.html http://postgresql.1045698.n5.nabble.com/tuning-on-ec2-td4341188.html http://aws.amazon.com/ec2/instance-types/ http://aws.amazon.com/about-aws/whats-new/2012/07/31/announcing-provisioned-iops-for-amazon-ebs/ https://gist.github.com/745635 http://aws.amazon.com/rds/faqs/ http://www.webyog.com/blog/2009/11/16/top-10-things-to-know-about-amazon-rds/ http://stackoverflow.com/questions/4806272/what-are-the-respective-advantages-limitations-of-amazon-rds-vs-ec2-with-mysql http://www.mysqlperformanceblog.com/2011/03/15/mysql-on-amazon-rds-part-1-insert-performance/ http://mysqlha.blogspot.com/2009/10/managed-mysql-amazon-rds.html http://www.mysqlperformanceblog.com/2011/02/21/death-match-ebs-versus-ssd-price-performance-and-qos/ http://aws.amazon.com/elasticache/ == Notes == * "CPU Cores" refer to virtual cores because hyperthreading is enabled ***** EBS-Optimized Instances: For a low, additional, hourly fee, customers can launch selected Amazon EC2 instances types as “EBS-Optimized” instances. EBS-Optimized instances enable Amazon EC2 instances to fully utilize the IOPS provisioned on an EBS volume. EBS-optimized instances deliver dedicated throughput between Amazon EC2 and Amazon EBS, with options between 500 Mbps and 1000 Mbps depending on the instance type used. When attached to EBS-Optimized instances, Provisioned IOPS volumes are designed to deliver within 10% of their provisioned performance 99.9% of the time. See Amazon EC2 Instance Types to find out more about instance types that can be launched as EBS-optimized instances. ***** == TO DO 1. Intro 2. tag cloud of amazon buzzwords 3. why the cloud 4. what they don't tell you. availability zones (HA) chaos monkey (link) http://status.aws.amazon.com/ elastic IP? 2012 ec2 outages *********** what we care about *********** ec2 Mi/S/(HC)Me/L/Me(HC)XL/HMXL/HMXXL/HMXXXXL/ ebs CloudFront RDS ElastiCache ************************ http://www.mysqlperformanceblog.com/2011/06/09/how-to-replace-a-ndb-node-on-ec2/ http://www.mysqlperformanceblog.com/2012/05/16/benchmarking-single-row-insert-performance-on-amazon-ec2/ http://www.mysqlperformanceblog.com/2010/06/17/high-availability-for-mysql-on-amazon-ec2-part-1-intro/ ************************ Ryan: kelp you did some writeup about the cost-benefit-analysis of ec2 some time ago, right? [10:50am] Ryan: do you have a link for that? [10:50am] kelp: I made a spread sheet of it at my old job [10:50am] kelp: but I don't have anything sharable [10:50am] Ryan: gotcha. [10:51am] kelp: but I could come up with something again if you need info [10:51am] Ryan: only if it's off the top of your head. i'm just curious [10:52am] kelp: at XXX we had about 25 server racks, and we're paying too much for our datacenter space [10:52am] kelp: buying that, and paying a couple of people to manage it [10:52am] kelp: still ended up cheaper after about a year [10:52am] kelp: and you keep the hardware about 3 years [10:53am] Ryan: the tipping point was like 30 racks? [10:53am] kelp: the tipping point is mostly just time [10:53am] kelp: I could probably work up another spread sheet [10:53am] kelp: to compare squares use [10:54am] Ryan: lol no chance we could use ec2 [10:54am] kelp: 1) we'd have to convince infosec [10:54am] kelp: 2) all the other reasons [10:56am] kelp: it would be really intersting to do the math with the SSD instances [10:56am] Ryan: haha yup. [11:00am] kelp: we'd almost certainly have to shard square/web [11:00am] kelp: largest instance is 60GB of RAM [11:06am] kelp: crazy, reserved pricing is SOOO much cheaper than on demand pricing [11:11am] kelp: with reserved pricing, aws isn't nearly as expensive as I thought [11:11am] kelp: I should really redo my model and see how it comes out [11:11am] kelp: the big issue is the high IO ones only exist in virginia and ireland, and we certainly need those [11:14am] kelp: ah, but we could do EBS volumes with provisioned IOPs in about any location