More EC2 Power

Amazon EC2 users now have access to a pair of new "High-CPU" instance types. The new instance types have proportionally more CPU power than memory, and are suitable for CPU-intensive applications. Here's what's now available:

The High-CPU Medium Instance is billed at $0.20 (20 cents) per hour. It features 1.7 GB of memory, 5 EC2 Compute Units (2 virtual cores with 2.5 EC2 Compute Units Each), and 350 GB of instance storage, all on a 32-bit platform.

The High-CPU Extra Large Instance is billed at $0.80 (80 cents) per hour. It features 7 GB of memory, 20 EC2 Compute Units (8 virtual cores with 2.5 EC2 Compute Units each), and 1,690 GB of instance storage, all on a 64-bit platform.

The AWS Simple Monthly Calculator now supports these new instance types.

We've been working with a number of tool vendors to line up early support for this important new feature. I plan to update the blog post several times in the coming days as this support becomes available.

-- Jeff;

Comments

[Steve Bjorg]

This is fantastic and a much welcome addition to the EC2 family. I know our customers will love it too!

Keep up the good work.

[M. David Peterson]

Oh, this is cool! :D

[Dan]

Excellent.

I'd also love to see "micro-CPU" instances. For example, I run a small instance with a whole bunch of miscellaneous tasks (monitoring, management web pages, nightly backups, flushing queues, etc). Even with them all running on the same box, my small instance typically uses less-than-100% of the resources available.

What I'd really like to is isolate each of these tasks on it's own instance. Then, one going crazy wouldn't take out the whole thing.

Ideally, a micro-instance would be about 1/10th the resources of a small instance.

These would also be great for prototyping: I'd like to test that things run across hundreds of machines before doing production deployments, but for early tests, it's a waste of money to allocation 100s of smalls...

Thanks!

[no@thanks.com]

Great news. While I understand the need for relative cpu units it would be useful to have some statistics available for the performance we can expect. Some standard benchmark results (yes this is difficult.. try defining "standard") for the processor and disk systems would be incredibly useful. For example, how a given extra large instance using a RAID setup across the 4 x 420 GB what is the max MB/s I can do when streaming sequential data into memory? Or how about using the ec2 persistent volumes? Can I aggregate a bunch of them and get high MB/s for scientific computing data sets?

[David Cancel]

Great news.

How about the inverse?

I need big memory low-cpu instances. Would think this to be a common pattern given the number of people using EC2 to host Apache web and Memcached servers.

Cheers,
David

[no@thanks.com]

Absolutely agree on the big memory instances. I would love 128GB or 256GB even. However, this does obviously clash with using virtualized infrastructure.

An alternative would be someone brave enough to deploy a software distributed shared memory solution on ec2 to aggregate several instances. I wonder if anyone has attempted to get an SSI like Kerrighed running?

[Vectorpedia(Rick)]

Thanks for keeping us informed on this new feature........

[Bruno Waes]

If only there would be EC2 in the european datacenters ... with free transfers to S3-EU.