We have added today 2 new Shared roles in the US East region, 32- and 64bit version of Ubuntu 10.04 pre-configured to run and scale MySQL version 5.1.41. They are both EBS-based, and use the new Scalarizr as the Scalr agent.
Other regions to come soon!
Not an Ubuntu fan? We just released 4 new pre-made roles in the US East region of EC2, with CentOS as the operating system. For those who don’t know what CentOS is, it’s a linux distribution that mirrors Red Hat, but without the trademarked material like logos.
This means that you can now scale your website (or individual components of your web stack) easily on an RPM package-based system. Two roles are available, ‘Base’ (which is ready to be customized into roles like Application servers, or ffmpeg transcoding servers) and ‘MySQL’ (which is ready to scale, using version 5.0.77). Both are EBS-based, which means that they can be start / stopped like any virtual machine, and run CentOS 5.4 (32 and 64bit).
US West, Europe, and Asia-Pacific regions to follow soon, as well as more flavors of linux.
These Roles use the all-new Scalarizr, our lightweight python agent that reports usage metrics back to Scalr to trigger the scaling. More on the Scalarizr in a later post.
Cheers,
The Scalr Team
Update: The new server’s IP address is 74.54.243.170, please add this to your security groups, with UDP ports 161-162 and TCP ports 22 and 3306.
Next Thursday (August 26) at 1:00 AM PST, we will add a new, non-cloud server to our infrastructure to improve fault-tolerance and performance. No UI downtime is expected, but there is a small chance that you might not be able to login a few times.
This is the last upgrade required to achieve full geographic fault-tolerance. Our nameservers are spread across 4 network-independent datacenters and two continents, and our http servers, polling servers, and database servers will all be fully redundant after the server addition.
Cheers,
The Scalr Team
Good day to all the Scalr users out there!
We added the set of pre-made Scalr roles to the Asia-Pacific region, so you can now scale in the most populated region of the world. This includes among others the NGINX load balancer, the Apache and Tomcat application servers, and MySQL database server.
You can set each of those to scale based on any of the Scalr metrics, such as Load Average (CPU), Free RAM available to the system, Network I/O, a fixed schedule, http request/response time, among others.
Congratulations to our friends at Rackspace and NASA Nebula, who launched OpenStack.org today!
Open Stack is an “open source, open standards” stack for Cloud Computing, which includes Compute (like EC2) and Storage (like S3). You can find all the code for Open Stack on Launchpad, as well as the other sub-projects like the iPad and iPhone apps.
This is a huge leap forward, probably the single most important announcement in the industry since EC2.
How does Scalr play into this?
We’re thinking about it. Scalr is currently licensed under the GPL v2, and we’d have to change that to the Apache License which the Open Stack charter requires. The Apache License precludes the dual licensing business model, in which we sell a commercial license to use Scalr for those averse to the GPL.
Open Stack also has a nascent Web UI, with a subset of Scalr’s functionality.
If anyone wants to weigh in here, their thoughts are welcome.
Today Scalr released the Scheduler, a Cron Job task manager designed for Cloud environments.
The problem is that Cron jobs, which are scripts that are executed periodically, were up until now tied to individual servers. When that individual server failed, the Cron job wouldn’t be executed. When that server was scaled, it would be executed once per server. This means that the tasks you set to be executed once per hour, could be executed 5 times in the hour, or not at all – not very reliable, to say the least.
Scalr based the Scheduler off Apache ZooKeeper, which is a part of Hadoop. It is a centralized service for providing group services, especially distributed lock services. The Scheduler extends Scalr’s scripting interface, and allows you to write scripts and set their execution schedule and scope: once on all servers, every hour on a subset of servers, daily on a single server, and any other combination!
As always, the code is available on Google Code and delivered as a service at scalr.net.
***
You can find the Scheduler under the Tools menu, found at the top when logged in:
You can schedule tasks to be executed periodically using the Scalr Scheduler
To add a Cron job or schedule a script, click the + icon next to the Search box:
Click the + icon to add a new cron job or to set a schedule for a script
You can then select the script to be executed, choose the parameters for it, schedule a time or interval, and run!
Configure the cron job's interval, parameters, and more
We integrated phpmyadmin, a database administration tool, into Scalr today. It allows you to see the data you have on your database, update individual records, alter tables, and do lots of other stuff.
This is another step into making Scalr a great tool for sysadmin work, and making great sysadmin more accessible to non-specialists.
You can find it under Farms > View, and from options, select MySQL status. Click ‘Setup PHPMyAdmin access’, wait a minute, refresh, and you’re done!
Let us know what you think!
If you used Scalr.net these last few days, you probably experienced some mysql connection errors. These were due to a very large amount of concurrent connections, typically when a large farm (100+ instances) is launched, as it triggers too many requests too quickly, before Scalr can react to load.
This was more of an architectural flaw, so we worked to reduce the amount of requests every instance makes. In many cases we got it down to a single request. For example, when an instance requests a list of instances of a role, it now gets all the information for all roles in the single initial request. Same goes for the config_opts queries from instances, equally optimized (we brought down the amount of requests to rebuild /etc/aws/hosts from 5 requests to a single one).
The next thing we did is tune mysql to handle thousands of connections, and over 100 other settings and sysctl options, then optimized our db structure (incl. added new indexes).
We also moved some stuff higher up in the stack to nginx, to be served faster.
We took the occasion to rewrite the client dashboard so that logs load instantly. You’ll notice this when you first log in.
Bottom line is that things are faster for you, and put less load on us.
Hi all,
Amazon announced availability of a couple new features last week, one of which is the possibility of using Elastic Load Balancing for EC2 in the EU region.
This will interest those who have users in Europe and have created a farm in the EU region to reduce latency. They can now save on a dedicated instance serving as a load balancer, as Scalr has added support for it.
Cheers,
Sebastian
Last week we announced Scalr Mission Critical, a service that guaranties that someone will be there to help if your site goes down. It gives you phone support, IM support, and short response times when using support.scalr.net.
This week we are proud to announce Scalr Development, a free version of Scalr open for developing your application while you don’t need the scalability requirements of a Production environment. This edition is limited to subdomains of development.scalr.net, so you can use myapp.development.scalr.net, for example. When you are ready to launch, just use ApacheBench or Tsung to test reaction to massive load, then sign up for Scalr Production and you’ll be good to go.
We are only accepting a limited amount of users this week, so please be patient if you are not accepted immediately.
Cheers,
Sebastian, on behalf of the Scalr Team.