Comments on: Short Comings of Amazon EC2

By: Ken Thomas

Ken Thomas — Mon, 01 Feb 2010 01:17:35 +0000

Kudos to Pavel, for already stating the obvious. Most of your points are seriously off, and you should at least consider updating the article to address this.

To the immediately above: you shouldn’t be using your root domain (versus http://www.example.com, etc, anyway :)

By: Son Nguyen

Son Nguyen — Mon, 18 Jan 2010 18:53:41 +0000

Still something to consider: Load Balancing still only supports CNAME entry, which has some severe consequences (those trying to point to the root domain). What EC2 needs is a standard A record to an elastic IP.

By the way, anyone has creative ideas on how to you scale down a cluster if there is existing data on the instance(s)?

By: Pavel P

Pavel P — Wed, 18 Feb 2009 10:39:47 +0000

Due to EC2 developments, most points currently can be considered wrong. I'd like to clariy them:

No IP address persistence (they all function as DHCP clients and are assigned an IP). One has to use dynamic DNS services for a given domain.

Obsolete. There is now. Elastic IP.

No block storage persistence. When the instance is gone, the data is gone. Yes I know you can send this back regularly to S3, but isn’t that actually a ‘hack’?

Obsolete. There is now - Elastic Block Storage. I've seen some articles claiming that it is rather good and quite fast.

No opportunity for hardware-based load balancing (which happens to be the key to scaling a process based framework like Rails and mentioned above).

Are we talking of sites with existing montlhy costs in tens of thousands $, or startups?

Most of "Web 2.0 Startups" are not able to afford hardware load balancing anyway. As for software-based balancing - there even might be ready-made images, just register and start it.

Also load-balancing is on AWS' roadmap.

No vertical scaling (you get a 1.7Ghz CPU and 1 GB of RAM, that’s it). So like the block storage problem, this hits databases, we run about 32GB of ours in memory.

Obsolete. There are different types of instances, some of them are quite good:

Extra Large Instance 15 GB of memory, 8 EC2 Compute Units (4 virtual cores with 2 EC2 Compute Units each), 1690 GB of instance storage, 64-bit platform
Large Instance 7.5 GB of memory, 4 EC2 Compute Units (2 virtual cores with 2 EC2 Compute Units each), 850 GB of instance storage, 64-bit platform

There are "extra CPU power" version of them too.

If you need even more processing power, you will be having your own datacentre anyway.

Can’t run your own kernel or make kernel modifications so there’s no ability for kernel and OS optimizations, and no guarantee that they’ve been done.

Any particular reason? Most VPS will not allow this too.

Images have to be uploaded and then moved around their network to find a launching point. This can take several minutes, if not more. Move 100 GBs around a busy gigabit network sometime and see.

Let me see. Last time when I brought 2U box to colocation facility, it took about 4 hours to get it up and running (security check-in, rack installation, setting up power, network ports, monitoring, testing, paperwork). Several minutes sounds fantastic!

Webserver image would be 1-2 GB, same for database image (basically, OS and application software). Application data would be on EBS volume. Images seem to be compressed, so free space is not "wasted".

What in the world would need 100 GB image to function?

By: Nitin

Nitin — Mon, 27 Oct 2008 07:18:49 +0000

I don’t know if carson is insane or not, but his points are invalid. Load balancing and persistent block storage are huge issues for a server applications, and Amazon is also aware of that. They now have persistent block storage with some really nice features, so at least they are not insane. I would love to see a GoGrid like load balancer.

By: Anonymous

Anonymous — Thu, 23 Oct 2008 17:28:08 +0000

EC2 just announced autoscaling, load balancing and monitoring.

http://aws.typepad.com/aws/2008/10/big-day-for-ec2.html

By: Son Nguyen

Son Nguyen — Thu, 09 Oct 2008 21:26:59 +0000

I agree on the problem with load balancing between EC2 instances. If EC2 has a built-in method to do this, it would be very nice and simple for users. Right now, there would need to deploy an instance doing software load balancing.

By: Matt

Matt — Fri, 22 Aug 2008 04:16:49 +0000

It looks like Amazon has just announced persistent storage for EC2? I’ve been trying to look around to see if there have been any reviews or feedback about how well it works.

By: Mohanjith

Mohanjith — Sun, 03 Feb 2008 00:45:59 +0000

For all of you concerned about persistent storage please read the blog post http://blog.mohanjith.net/2008/02/amazon-ec2-with-rock-solid-persistent.html. I was able to get harware node level data persistence. It’s worth looking at.

By: Ralph

Ralph — Wed, 23 Jan 2008 22:13:01 +0000

My issue with all of these services is how do you get a physical backup of your database? I’d love to use these SimpleDB, for instance…but I would really be into it if you could pay for like a BluRay or tape backup to be mailed to you every 2 weeks or something. Otherwise, I’m kind of iffy on the whole thing.

By: Tom

Tom — Mon, 14 Jan 2008 01:52:08 +0000

Amazon now have a database specific web service, which probably negates the need to use EC2 to spin up a DB server. It seems like it would be more scalable too. I guess unless you really need a specific extension to mySQL or something then you should be fine using that.