Red Hat projects to seed cloudy IT
Let's get one thing straight. We don't like the term cloud computing any more than you do Of course, Richard Stallman doesn't like when we call it Linux rather than GNU/Linux. He's gotta live with Linux. And, well, we've gotta live with cloud computing. It's not going away. Commercial Linux and middleware distributor Red Hat is …
"the company can't sell software because that violates open source licensing". WTF? Selling open source software doesn't violate ANY major open source license, certainly not the GPL. Selling open source software is precisely what Redhat and every major open source company does.
Such a complete lack of industry knowledge, coming from a so-called journalist is shocking.
In fact, my understanding is that any license that forbade selling the software, would violate the first clause of the open source definition, (although it does require that the software is also available, at least in source form, free gratis):
"1. Free Redistribution - The license shall not restrict any party from selling or giving away the software as a component of an aggregate software distribution containing programs from several different sources. The license shall not require a royalty or other fee for such sale."
What RedHat are mainly selling, though, is support for the software they distribute, and this is why people pay for it (rather than downloading essentially the same software at no charge from CentOS).
I can already do most of this in our enterprise environment with Puppet and Cobbler...
Voldemort, Cassandra, and MongoDB are not cloud filesystems, or file systems of any kind for that matter. They are structured data-stores, aka databases. Mentioning these projects in this context is misleading and likely to confuse many people.
Can I correct the claim in this article that LinkedIn's Project Voldemort and Facebook's Cassandra are filesystems. They are column-table database thingys,. not filestores. They may use the Hadoop Filesystem APIs, but those APIs are designed to work with multiple filesystems, not just Hadoop's own HDFS.
I don't see why RHEL has to rewrite the entire cloud stack from the ground up. Do they really know better than facebook and Yahoo! what a datacentre-scale application needs? How many datacentres with 10+ Petabyte filestores do they have?
No, Project Voldemort (http://project-voldemort.com) is a distributed key-value store, and Apache Cassandra (http://incubator.apache.org/cassandra/") is a column family oriented database. Neither of them uses the Hadoop Filesystem APIs.
Their idea of the in memory file system is pretty bad. I mean google was doing that what 10+ years ago? Data sets have grown well beyond what will effectively fit into memory hence large buildouts of server farms with cheaper disks and distributed computing that way.
If anything I would of thought they would stick to disk based systems and at least go with SSD, maybe FusionIO or something, get near memory performance and get massive amounts of space at the same time, I mean when was the last time you saw a server with 300+GB of ram on it?
And as for cluster file system I can't imagine they wouldn't build off of what they have in GFS2 already..
