Cheap Network Attached Storage Is Expensive.

The bottom line: you get what you pay for…

In my limited IT experience (~20 years), I have seen it proven over and over that low-cost hardware costs a lot in the not so long run. This post is about NAS, but this premise is not limited to that particular device. In the course of various consulting engagements, I have come across many devices that on first glance seem to be a great value. Until you really get it in production…

Before I start making sweeping generalizations, let me be clear: many manufacturers have a wide breadth of products, many of which are truly commercial-grade, enterprise ready. This is not what the common small business or non-profit chooses the first time around. For example, Iomega delivers several wonderful, relatively low-cost devices that work very well as a disk-disk backup target. The ix4-200 comes to mind.

The real problem comes from the low end hardware in the low-cost NAS devices. This includes home PC quality hard drives, incorrectly sized power supplies, improper cooling, lack of standard storage protocols available, no redundant drives, no hot-spare, and generally poor user management interfaces. A “good” device will provide a non-proprietary way to access the data if the rest of the network (servers/PCs) fail.

When purchasing any type of network storage (or rolling your own), carefully consider all of the components before making a purchase. Look at the MTBF (mean time between failure) or CDL (component design life) of the drives. The difference between a true RAID-quality drive and a PC drive can be staggering. Remember that your NAS runs 24/7-365. That Is 8760 hours per year. Often the better equipment is only a few dollars more and will save you considerably less in the long-run.

How much is your company’s data worth to you? Industry statistics have shown that 60% of companies that lose their data shut down within six months. Over 90% of companies that lost their data center for ten days or more file for bankruptcy within one year. Every week there are over 14,000 hard drive crashes in the United States. Don’t let your company become a statistic…

– habanero_Joe

Open Source NAS – Openfiler Update

Open Source Storage
The online debate over Open Source storage options is fast and furious. The two common names that keep popping up are FreeNAS and Openfiler. Do a Twitter search for either and you will have an afternoon full of reading.
I evaluated Openfiler (see April 7 post for Openfiler info) and have not looked back. The first box I installed as NFS storage has been running for 54 days with no data loss or corruption. The only issue at all has been the failure of one of the on-board Ethernet ports. My second box installed as iSCSI storage has been in production for thirteen days.

Hardware breakdown (retasked servers – nothing new was purchased):

filer1: Dell PowerEdge 2650, dual Intel Dual-Core Xeon 2.4 Ghz CPU w/512KB cache, 4GB RAM, dual on-board Gigabit Ethernet ports, (1) Maxtor Atlas 10k 74GB harddisk for Openfiler system, (4) Seagate Cheetah 10k 146GB harddisks for shared storage.

filer2: Dell PowerEdge 2850, dual Intel Quad-Core Xeon 2.8 Ghz CPU w/2.00MB cache, 4GB RAM, dual on-board Gigabit Ethernet ports, dual-port Intel PRO 1000 NIC (for iSCSI), (2) Maxtor 15k 74GB harddisks in RAID 1 for Openfiler system, (3) Seagate 15k 146GB and (1) Fujitsu 15k 146GB harddisks for shared storage.

Filer1 reports used physical memory of 97% while filer2 reports 8% physical memory in use. This is interesting as filer2 has a much higer load at this point. IRC chats, blogs and forums suggest Openfiler likes all the RAM it can get its hands on.

Plan Moving Forward
I will continue to use filer2 as iSCSI storage for two VMware ESX 3.5 servers. Planning to rebuild filer1 with additional RAM and add Intel PRO 1000 card. May convert storage to iSCSI for Citrix XenServer evaluation.