When RAID is not equal to High Availability

Know your storage devices before buying something “just as good”.

An SMB experience using low-cost iSCSI storage devices

One objection we typically hear from potential clients is why vendor X costs more than vendor Y for the same features and specifications. Today, it is not uncommon in the storage market (SANs and NAS) to know that that you are buying more than specs and need to look more in-depth when you find a low-cost alternative for a technology that previously was considered too expensive for your organization.

My recent client experience exemplifies this problem. The storage needs have grown over the past several years and exceeded their file server capacity. Their business requires the long-term retention of image files for every device they manufacture and the repository is approaching 25 years-old. Many of these image files may never be retrieved unless a customer requests them or engineering is researching a design or manufacturing defect. Their daily requirement is to capture high-resolution images of their products and continuously store them as they are prepared for final shipment.

The annual IT budget planning always includes a small SAN for the long term storage and daily storage of these important image files. However, the first item to typically be removed from the budget is the SAN due to what is considered the high cost of the product. Last year this changed, with the introduction of a low-cost, high-value storage device from Buffalo Technology. Who would not want an 8TB rack mount iSCSI RAID5 storage device for $2500 that is “just as good” as small iSCSI SAN for $15,000? The IT budget could surely accommodate such a low-cost, high-value item.

The first year was uneventful and the device performed as advertised. Recently, the device configured with four 2TB SATA drives indicated a drive failure in Slot1. This should not be a problem, as we all know RAID5 will keep running with a single failed drive. A call to Buffalo Technology and the completion of some basic troubleshooting confirmed the problem. A new hard drive replaced under warranty would solve the problem. After jumping through several hoops, chasing confirmation emails, faxing in receipts and paying $100 for an advance replacement my new drive showed up six days later. Yes, I could have paid more for next-day shipping, after the receipt was faxed in, validated by customer service, and payment was submitted to their website. I suspect next-day shipping would still require at least three to four days for the drive to show up.

The instructions delivered with the drive were minimal at best. Update firmware on the device, and insert the new drive. Why would I want to update firmware on an iSCSI storage device that already has a failed hard drive is not something I want to risk and potentially lose all the data? Things happen in IT we all can attest to.

The new drive was inserted, the red LED started to blink and I waited, and waited, and it appeared to be rebuilding the array. I came back after a weekend and it was still blinking. I could not see progress on the rebuild and the product manual gave no indication of the actual process, but being an IT professional I have worked with many RAID5 storage devices that when a failed drive is detected and removed, and a new drive is inserted it just starts rebuilding and your done after several hours.

Not this device, a call to support confirms that you need to go into the web interface, detect the new drive, and then select some menu items to initiate the rebuild.That did not go as planned, and support had to check with a more knowledgeable resource. After a brief wait on the phone I was told to dismount the iSCSI storage device and start the rebuild again. I responded that dismounting the iSCSI device would take my storage offline and make in unavailable to my network of users. Why would this be a requirement for a RAID5 iSCSI storage device used by business? I followed their recommendations and then discovered it would be 33 hours for the array to rebuild with the new drive. I now had a client that was not happy that their RAID5 iSCSI storage device that was “just as good” as a more expensive iSCSI SAN would require them to stop using it for 33 hours while the array was rebuilt. The final outcome was that the unit was restored and all data stayed intact, however, the experience with a product that was “just as good” was much less than expected.

Paul R. Cook
Paul R. Cook
Vice President, Network Services Group
Paul.Cook@CustomSystems.com

 

 

© Copyright 2014 Custom Systems Corporation

Leave a Reply

Your email address will not be published. Required fields are marked *