Rackspace launches new OpenStack cloud | Hacker News Reader

Rackspace launches new OpenStack cloud

rackspace.com

63 pointsteoruiz22 comments

This is actually really great. One of the things I hated so far about Rackspace Cloud was that I couldn't scale the resources independently. If I need to upgrade to multiple CPU's or more RAM I don't want to be required to double or triple my disk capacity - that's a giant waste, especially with the availability of Cloud Files.

Now if their Database hosting would include Mongo, I'm sold!

Judging from the price page, it doesn't actually look like this does that. The price tiers are still some combination of a fixed amount of storage and RAM per tier, with no mention of number of cores. Maybe this is because their EBS clone isn't ready, yet?

Cloud Block Storage is available through the limited availability program and it is billed as a separate add-on to Cloud Servers so you can have a flexible amount of additional storage to go with the fixed amount provided with each CPU/Memory instance type.

[edit added to answer the reply] This is not the dNAS solution available as part of our dedicated cloud platform offerings that can be connected to Cloud Servers over RackConnect -- it is an API accessible block storage solution that is part of our public cloud platform. [/end edit]

Is that their dNAS product or something else?

This seems to address the use case of wanting more storage for a given amount of memory, but not the reverse. You can't, say, get 16gigs of ram, without also getting a 620gig disk, nor can you control core count independently of the other variables, in contrast to (for example) EC2, where there are a variety of instance types with different tradeoffs in CPU and memory, and storage is effectively always added (and billed) separately. It's unfortunate that Rackspace isn't taking advantage of the opportunity this relaunch presents to increase the flexibility of their product.

Can you mount the storage on your other servers?

As I remember rackspace is a cloud in the same sense that Linode is a cloud, but not in the same sense that Amazon is a cloud, ie. they've taken an existing vps product (Slicehost) and resold it under the 'cloud' brand rather then developing something new from the ground up as a cloud.

Per the blog post, Cloud Servers is no longer based upon an existing VPS product. Instead, it is based upon the OpenStack project that Rackspace started with NASA in 2010. http://en.wikipedia.org/wiki/OpenStack

They use scalr api so that you can upgrade as you need it.

This is my (only) beef with Rackspace. We're currently using them, but we're looking to migrate away. Because of their tiered pricing, we need to pay through the nose for our ssl-proxies/edge caches, since that's the only way we can get enough bandwidth on them.

I'm glad and grateful they've open sourced their stack.

Yet it sheds some amusing light on their pricing;

  Openstack at Rackspace: 30G Ram, 1.2T Disk = $1314/mo

Rent a box at Hetzner, take the time to install Openstack;

  Openstack at Hetzner: 32G Ram, 2x3T Disks, probably faster CPU = $77/mo

Obviously that only makes sense when your time is "free", and Hetzner is not exactly as solid as Rackspace. And you should rent a couple more boxes for redundancy.

But still, I'm almost tempted to start a little Hetzner-Cloud...

It's good to see the option of SSD-backed block storage in a major cloud provider like Rackspace. Block IO is a killer for a lot of apps. Though, we'll have to see if the benchmarks support their claims though.

Where did you see the mention of SSDs?

Scroll down to their diagram and click on "Block Storage" on the left. It mentions the choice of SSDs or SATA drives.

I think hackernews is overloading the cloud. I can't load the video.

"OpenStack® is the open alternative to proprietary cloud platforms and lock-in."

Oh the i®ony.

There are many registered trademarks in the open source world, including Linux, Ubuntu, and Wikipedia among others. OpenStack is also a registered trademark. I don't see any irony at all. Registering a trademark is a way to protect an open source project from abuse by people who would take advantage of the name to endorse their own unrelated product. Try getting the source code for aws and try getting the source code for open stack, and then tell me which one you think is more open.

The trademark will transfer to the foundation. It's just a way to ensure that anyone claiming to run an OpenStack compatible cloud is indeed doing just that. It's good for an OSS community to trademark brand assets and protect the name and meaning of their project.

Couple of additional comments about the OpenStack projects all of the new services are powered by on this HN submission: http://news.ycombinator.com/item?id=3847486

Here's to hoping that its cloudfiles implementation no longer uses a file replicated sqlite db for indexing. (iirc it did earlier on. Scared me stupid when I found out).

Can you explain why the account and container index architecture in OpenStack Swift (which powers Cloud Files) scares you? Have you really looked into the architecture or did you just hear the information anecdotally? A link to the architecture overview: http://swift.openstack.org/overview_architecture.html

I don't see any information on the CPU resources available or IPv6 support.

Reply on news.ycombinator.com