Home page logo

nanog logo nanog mailing list archives

Re: Reliable Cloud host ?
From: Jon Lewis <jlewis () lewis org>
Date: Sun, 26 Feb 2012 19:29:40 -0500 (EST)

On Sun, 26 Feb 2012, Randy Carpenter wrote:

This does not have to be true at all. Even having a fully fault-tolerant SAN in addition to spare servers should not cost much more than having separate RAID arrays inside each of the server, when you are talking about 1,000s of server (which Rackspace certainly has)

When is your cloud offering going to be available to the public?

I don't need that kind of HA, and understand that it is not going to be available. 15 minutes of downtime is fine. 6 hours is completely unacceptable, and it false advertising to say you have a "Cloud" service, and then have the realization that you could have *indefinite* downtime.

I think you're assuming "cloud" means things that the provider does not. To me, "cloud" just means VPS that I can create/destroy quickly whenever I feel like it, without any interaction from the provider's people. i.e. a few mouse clicks or an API call can "provision a 10gb CentOS VM with 256mb RAM". It's up and running before I could locate a CentOS install CD, and if I don't like it, a few clicks or an API call deletes it, reprovisions it, exchanges it for a Ubuntu server, etc. Cloud doesn't mean if the node my VM(s) are on dies or crashes, my VMs boot up on an alternate node. That would certainly be a nice feature, but that's just a form of redundancy in a cloud...not a defining attribute of cloud.

The funky problem with DNS specifically, is that all the servers need to be up, or someone will get bad answers. Not having a preference system,

DNS "handles" down servers. How would one of your DNS servers being down give someone bad answers? It won't give any answers, and another server will be queried. Or do you mean if storage "goes away", but your DNS server is still running, it'll either give nxdomain or stale data...depending on whether it had the data in memory or storage went away and updates began failing because of it?

 Jon Lewis, MCP :)           |  I route
 Senior Network Engineer     |  therefore you are
 Atlantic Net                |
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________

  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]