6 ms·

New EBS (AWS) volume type: General Purpose (SSD)

14 points by jacobscott 12y ago

It looks like Amazon has released a third EBS volume type, and renamed "Standard" volumes to "Magnetic". General Purpose is currently the default type when launching "Create Volume" in the EC2 Web UI (https://console.aws.amazon.com/ec2/v2/home), which provides the following documentation: "General Purpose (SSD) volumes provide the ability to burst to 3,000 IOPS per volume, independent of volume size, to meet the performance needs of most applications and also deliver a consistent baseline of 3 IOPS/GiB. Provisioned IOPS (SSD) volumes can deliver up to 4000 IOPS and are best for EBS-optimized instances. Magnetic volumes, previously called 'standard volumes', deliver approximately 100 IOPS on average, with a best effort ability to burst to hundreds of IOPS." The EBS Pricing page (http://aws.amazon.com/ebs/pricing/) currently makes no mention of General Purpose (SSD) volumes.

exidy 12y ago
Interesting. I can create a gp2 volume through the console, but not through CloudFormation. UPDATE_FAILED - gp2 is invalid. Valid volume types are standard and io1.
- dandekar 12y ago
  We can look into this if you can provide the stack id.
  - exidy 12y ago
    Thanks, but I'll contact AWS support through our account.
- matt_oriordan 12y ago
  See https://forums.aws.amazon.com/thread.jspa?threadID=154543&tstart=0&messageID=550673 https://forums.aws.amazon.com/thread.jspa?threadID=154543&ts.... For some bizarre reason the new SSD types are not supported with Autoscaling groups.
nnx 12y ago
Official announcement just been released. http://aws.amazon.com/blogs/aws/new-ssd-backed-elastic-block-storage/ http://aws.amazon.com/blogs/aws/new-ssd-backed-elastic-block...
albanr 12y ago
It'll be interesting to see how much CloudCorset will be able to impact/improve performance by reducing the number of IOPS that goes to the EBS storage backend. http://www.cloudcorset.com/?ref=Tie98 http://www.cloudcorset.com/?ref=Tie98
renaudg 12y ago
Is it supposed to be all smooth sailing launching gp2 instances from standard snapshots and vice versa ? It may be just me but I've had really weird and serious new errors today in us-east-1 launching new instances, or even starting previously stopped ones with no changes to the instance whatsoever ! General slowness, "reachability" health check failing, with the system log showing that the kernel couldn't be uncompressed, maybe because the bootloader couldn't find the root device in the first place ? I'm not using custom kernels at all. The timing is too strange for it to be a coincidence, but nothing on the AWS status page.

14 points by jacobscott 12y ago

exidy 12y ago

Interesting. I can create a gp2 volume through the console, but not through CloudFormation. UPDATE_FAILED - gp2 is invalid. Valid volume types are standard and io1.

dandekar 12y ago
We can look into this if you can provide the stack id.
- exidy 12y ago
  Thanks, but I'll contact AWS support through our account.
matt_oriordan 12y ago
See https://forums.aws.amazon.com/thread.jspa?threadID=154543&tstart=0&messageID=550673 https://forums.aws.amazon.com/thread.jspa?threadID=154543&ts.... For some bizarre reason the new SSD types are not supported with Autoscaling groups.

nnx 12y ago

Official announcement just been released. http://aws.amazon.com/blogs/aws/new-ssd-backed-elastic-block-storage/ http://aws.amazon.com/blogs/aws/new-ssd-backed-elastic-block...

albanr 12y ago

It'll be interesting to see how much CloudCorset will be able to impact/improve performance by reducing the number of IOPS that goes to the EBS storage backend. http://www.cloudcorset.com/?ref=Tie98 http://www.cloudcorset.com/?ref=Tie98

renaudg 12y ago

Is it supposed to be all smooth sailing launching gp2 instances from standard snapshots and vice versa ? It may be just me but I've had really weird and serious new errors today in us-east-1 launching new instances, or even starting previously stopped ones with no changes to the instance whatsoever ! General slowness, "reachability" health check failing, with the system log showing that the kernel couldn't be uncompressed, maybe because the bootloader couldn't find the root device in the first place ? I'm not using custom kernels at all. The timing is too strange for it to be a coincidence, but nothing on the AWS status page.