Organisations such as retail chains, schools, hospitals, insurance companies, and many other businesses with remote locations face unique challenges when it comes to a storage solution. Provisioning multiple sites with storage to meet the local needs of each location, while ensuring that the storage can be managed and protected from a central site, requires a comprehensive approach with a scalable storage solution. This Q&A; discusses the top considerations storage administrators need to keep in mind when selecting and installing such a distributed NAS solution.
Assume that your company has multiple remote sites, each of which required NAS storage.
* The storage will be used by a heterogeneous network of computers running a variety of operating systems including Windows, Mac OS X, Linux, VMware, and UNIX.
* The sites are connected by VPNs over the Internet.
* The remote sites have little or no IT personnel, so the distributed NAS system must be capable of running continuously and reliably, while administration and data protection takes place at the central office.
1. Right-sized storage for each site
Each of your locations will have very different storage needs. Smaller offices will likely have very modest storage needs, while other sites may need a large amount of storage. Take some time to determine your needs first, to be sure you will have enough, yet not purchase more than you will need.
NAS storage typically holds primary data for individuals, groups, and key applications. To ascertain how much storage you will need to back up your local NAS system at the site, determine the amount of storage used by those computers and multiply that amount by 1,5 to 2,0 to conclude how much storage is needed to contain your backup data.
The total storage required is equal to your primary storage needs plus the additional storage required for backups. Storage needs often grow by 80% per year, so be sure to plan for your future needs, as well as what is required today.
Once you have determined your storage needs for each site, purchase the storage system that best fits the needs of each location, being careful not to over purchase. Since the initial purchase price of a distributed NAS solution is heavily dependent upon the size of the systems you buy, finding a vendor that offers systems in a variety of sizes - from small desktop units for your smaller locations, to scalable rack mount systems for your central office - is key to obtaining the storage you need, without breaking your budget.
Consider the physical space and environment of each location. Storage devices can be extremely loud during the course of normal operations, which can become a serious nuisance in a small office that lacks the ability to isolate it in a separate room. Some vendors make desktop storage units that are whisper quiet and are therefore better suited for these sites.
For a larger office with a separate server room, a 19-inch rack-mount storage may be more appropriate, since these systems have a great deal more storage capacity than the desktop units. And since they will be in a server room, noise will not be an issue.
3. Centralised management
One of the biggest advantages of a distributed NAS system is the fact that it can be administered from one central location, without the need for specialised IT personnel at each site. Once the NAS storage is installed at each location, you will be able to see the health of each system from a computer at the central office. If a particular NAS system needs attention, you will see it in your console and have the capability to administer it from the central office. Ask your NAS vendor to explain the steps required to get all the NAS systems securely communicating with one other and appearing on a single management console.
It is also good to configure each NAS system to alert key individuals when they are in need of attention.
NAS systems that are kept isolated drive up IT operational costs and significantly increase operational risk.
4. User identities
Each file and folder on the NAS system must be secured so that only specific individuals and/or groups will have access to them. Creating unique user identities that only exist on the NAS system can lead to mass confusion by users and create additional burden on IT personnel. Be sure your NAS system can integrate with a Windows Domain to use Active Directory user identities and/or use the user identities in a NIS directory service. The NAS system typically maintains continuous communication with the directory service to verify the identity of users.
Once they have been authenticated by the directory service, the NAS system will allow or deny access to its files and folders.
5. Data replication
The secure replication of data between sites is an essential component to your data protection strategy.
Replicating business-critical data from each of your remote sites to the central office ensures that nothing is left to chance. The essential data from each site can be treated consistently by qualified IT personnel in accordance with your established backup and recovery policies.
Replication software should run directly on the NAS system to avoid having to purchase and maintain a replication server. Data must be able to be moved securely, either through the encrypted channel provided by the VPN hardware, or by encrypting the data with software built into the replication software.
Since software-based encryption can potentially be a performance drain, the administrator should have the ability to set the amount of bandwidth utilised by the replication software. This removes the need for the Internet connection to be dedicated to that function replication should have the capability to either be run continuously, or on set schedules. Depending on the application, one-to-one, one-to-many, or many-to-one replication is appropriate. Lastly, larger servers in remote sites should have the capability to directly replicate to NAS systems in the central site when appropriate. Following the initial replication activity, only the bytes that change should be replicated.
6. Link speed matters
All too frequently, the speed of a site’s incoming telecommunication link is frustratingly slow. A T1 line is most commonly used (T1 leased lines offer the same data rate as symmetric DSL), but that only provides a maximum throughput of 1,544 megabits per second, which translates to a maximum speed of 1 GB per hour. Conversely, a T3 line provides 28x the maximum throughput of a T1 - but this option is cost prohibitive for many sites.
7. NAS system speed matters
It is often thought that performance is less important if only a few users are casually using a NAS system.
However, consider the data that resides on those systems. If a distributed NAS system is housing business-critical data, rapid retrieval will be essential.
Similarly, if the NAS system is part of a backup-to-disk solution, it must be capable of absorbing large bursts of data during the backup window. Lastly, if multisite management or replication software will be employed, as recommended, a high performance NAS system will be required to run them. Smaller NAS systems are typically not capable of delivering this level of performance.
A thorough data protection plan can only protect the data that existed at the time the last backup copy was made. To protect your vital data, ensure that each system in your distributed NAS configuration is capable of an acceptable level of RAID protection.
NAS systems can also use higher-quality commercial drives which have longer MTBFs and cycle times.
Redundant power supplies and fans are also common on larger NAS systems. Lastly, redundant LAN ports ensure the NAS system can be accessed if a port, switch, or LAN cable should fail.
9. Virus protection
Users who are not current in their virus protection software put other users at risk. NAS systems should run virus protection software to ensure the data it will house is virus-free.
10. Touch-free branch office data protection
A distributed NAS solution should be capable of protecting the Windows, Linux, Mac OS X, and UNIX computers that exist at each site. All data protection operations should be able to be performed from a central location. Long term archives should be created, and fast recoveries from local disasters should be provided. Backup and restore software should be used in combination with replication software to protect all servers, desktops, and notebooks.