For techie tips and tricks, tools and sites of (dis)interest

Storage solutions

| Sunday, December 05, 2010

External HDDs and Direct-attached Storage:

Given that many, if not most, of us consumers no longer use desktop PCs in their home, it's rare that we have more than 500GB of internal storage at our disposal. Online storage can provide us with somewhere to store our photos, and even documents and code, but this only adds a few GB (unless you opt for a paid service). As a result, the majority of users who need more storage opt for externall HDDs. They are a good short-term solution, those with eSATA connectivity are fast enough to run VM images, and a couple of 1.5TB disks can keep you going for a while. If you're a hoarder (like me), and/or you want some sort of redundancy, this isn't going to be enough.
Any old PC (not too old, though) can act as a decent file server, with minimal effort. I have an old Athlon that serves as a kind of DAS file server - not dedicated to the task, but it serves my modest needs and beats dragging external HDDs around the house. There are a couple of good NFS apps (or you can use SMB): FreeNAS (based on FreeBSD), and MS' own Windows NFS too.

Networked Storage - NAS and SAN:

Network-Attached Storage has been around a long time in small to medium sized businesses and corporate environments, but more recently it has become more popular with consumers too. It's possible to set up a NAS (using the NFS or SMB protocol), and is one way of expanding your external storage while making it accessible from anywhere on the network. However, ready-to-go NAS devices, ranging from ones nearly as small and portable as an standard external HDD, and not much more expensive, to racks, are completely self-contained: they have their own processor and embedded Linux. The most basic NAS enclosures can be picked up at a very reasonable price from resellers (well known brands available for under €80), and will network any SATA drive. Larger, more expensive NAS devices can network many more drives.

SANs are different to NAS, in that they don't provide any file system, but only storage. The way this storage is utilised is up to the client. SANs use protocols like fibre channel, and also iSCSI and ATA over Ethernet. The last two are a particularly good choice because they don't require dedicated channels of copper or more expensive optical fibre, but use the existing (and ubiquitous) ethernet network and a clever protocol that extends the normally local storage protocol over the network via IP. I don't really understand the ins-and-outs of the SAN controller, but once set up properly and well tuned, SAN performance should be indistinguishable from local storage.
For this reason it's an extremely popular enterprise solution. Many of the customers I deal with use it for storage for their application servers, DBs, LDAPs, mail servers and so on, and it rarely comes under suspicion these days (although it often did up until two or three years ago, so my impression is that they have become very reliable as time has passed).

Scalable storage solutions:

While combining storage with IP can allow you to share storage across a LAN or WAN, things can get messy as you try to scale up the system. Projects to upgrade or expand an existing SAN can be a real headache, and at the very least can involve a very stressful weekend for the IT team.
Using grid architectures, it's possible to build a small storage system first, and then expand it over time. Grid technology also provides redundancy and means upgrades can be carried out while the system is still live.
Enhanced by Zemanta

0 comments:

Post a Comment

speak your mind, but keep it clean (the comment, not your mind).
no spam and no trolls please