Exclusive: NetApp NearStore R150 makes server backup nearly plug-and-play

Network Appliance’s NearStore R150 storage server can solve a real problem that many datacenters face — even if they don’t realize they have it yet: How do you backup, archive, replicate, and protect the terabytes of data passing through databases, mail servers, and transaction systems? Mirroring servers requires duplicating expensive processors and applications, and doing so won’t protect against tampering. Using tape or optical is slow and not suitable for temporary backups. The R150 presents a practical and attractive option: Simply copy the data over the LAN to a huge network-attached file server that’s optimized for that purpose and dedicated to making highly secured file backups.

The product’s brand name, NearStore, reflects its intended purpose. “Nearline” is becoming a popular term to define a storage system that isn’t offline, such as data archived onto tape, but isn’t online in the sense of being directly accessible by applications or users. A nearline server such as the R150, therefore, is connected to the network, and its data is nearby in case it’s needed for restoring a damaged server; rolling back a system to an earlier data state; or producing records for an audit or subpoena.

I found the NearStore R150 server to be somewhat complex to set up and configure, but easy to run in a production environment. NetApp had predefined access to a FAS960 file server, which the company loaned the InfoWorld Test Center for that purpose, and the R150 was back to back it up immediately. However, the process was much more cumbersome when it came to defining my own data sources to be backed up, and setting up access to those sources, schedules, backup policies, archive rules, and so on. Part of that difficulty resulted from the need to maintain a secure computing environment: You can’t simply open up servers on your network so the R150 can use them.

However, I think NetApp has done all it can to make that process easy, including decent documentation and good user interfaces. The good news: Once those data sources are defined and have been successfully backed up to the R150 once, the system is close to “install and forget,” doing its work with minimal administration. The benefit is that a company won’t need to hire additional administrators to run a NearStore R150.

The full R150 system comprises a single R150 storage appliance, which Network Appliance calls the R150 head, as well as a set of dedicated disks, a rack, and software licenses for the server and for backup clients. The configuration NetApp supplied is the largest configuration the company currently offers, with 24 terabytes of disk storage.

The R150 head outwardly resembles one of NetApp’s FAS series file-server appliances, and frankly, that’s what I suspect is behind the all-black bezel. The 6U server head contains 6GB of ECC (error-checking and correcting) RAM, as well as six PCI-X slots, two of which are available for expansion.

The server, which runs NetApp’s proprietary Data OnTap operating system, came equipped with a dual-port GbE (Gigabit Ethernet) fiber card and a dual-port GbE copper card, both from Intel, as well as an extra 10/100 port, presumably for out-of-band management. The company offers optional Fibre Channel cards for connecting the R150 to a SAN, as well as SCSI cards to hook up to external tape libraries and implement hierarchical storage archiving.

For storage, this version of the R150 contained nine 4U disk shelves, each of which held 12 hard disks. Those disks were all 212GB ATA-based hard drives, which means lots of capacity at a lower price than comparable SCSI or Fibre Channel disks. Given the intended use of the R150 system, using ATA drives to reduce price is acceptable. Total raw capacity of the system was 24TB, the same figure Network Appliance promotes. The default net capacity, using RAID striping, is 21TB, athough that’s an administrator-selectable setting. Other RAID settings would result in greater data protection but reduced capacity.

The system has only minimal protection against hardware failure. The server head and disk shelves each contain dual hot-swappable power supplies and dual fans, and the devices are designed to go to separate electrical circuits. However, there is no redundancy in terms of the server head, so if it goes down, data won’t be backed up and archives won’t be accessible until the head is repaired or replaced by NetApp. There’s little in the system that’s user serviceable, beyond swapping out a hard drive, power supply, or fan.

Frankly, I would have felt better about the protection offered by the R150 if it had dual, cross-connected filer heads, and if it were able to strip disk arrays across separate disk shelves.

The value of the R150 isn’t in its hardware, which is generally similar to network-attached file servers from NetApp and other vendors. Rather, the proof of this pudding is in its applications software, SnapVault and SnapLock, both of which were included in our test configuration.

SnapVault is the main data backup utility, which can copy from external servers to the R150 server. The company provided two versions of the SnapVault software, one that copies software from NetApp’s FAS file-server appliances, and the other, which copies data from Linux, Unix, and Windows-based file servers.

SnapLock is designed to create WORM (write-once read-many) archives of data stored on an external server. The WORM aspect is to prevent key archives from being tampered with, in accordance with either regulatory or enterprise data policies. For regulatory applications, the software can be configured to prevent archives from being deleted at all; and NetApp offers other software, called SnapMirror, which can be used to move archives to secondary backup sources or to optical or tape storage.

In my tests, the NetApp system performed exactly as expected and required. Given my earlier experience with NetApp’s previous-generation servers and command-line tools, I was expecting a system that would be hard to administer and manage, but the company has made strides to improve its software usability.

Today, NetApp offers a Web-based interface that allocates so-called “secondary volumes” to store file archives. That same user interface is used to assign source servers to those secondary volumes, configure scheduled backups of those volumes, and set retention policies. The company even offers some wizards for setting up backups and restores. Plus, the R150 has to be given read access to the backup source; restoring data, of course, requires read/write access.

The SnapLock software is particularly impressive given the sensitive nature of archived documents, especially in light of recent regulatory and financial scandals. The idea that a backup onto a file server can be nearly as well protected as backups to a secure off-site or optical vault is impressive. Of course, if miscreants have physical access to the R150 system or if the server is destroyed in a disaster, the data could still be in peril, although the files themselves should be tamperproof. Thus, someone concerned about the data should look at the SnapMirror option to add an extra layer of complexity.

Another suggestion for anyone looking at the R150 solution: Consider creating a secure second Ethernet LAN to connect the R150 and its source servers. That second LAN would make it less likely that the data backups would impact production systems, and it also offers an additional layer of data protection. But even on a primary datacenter LAN, the NearStore R150 solution is strongly recommended as a data-protection solution.