This post has been written by Brian Flora, a writer and a self profound tech geek, who specializes in data storage that includes cloud computing and flash arrays. This is what he has to say about the best practices for storage provisioning in enterprise application deployment.
Considering that data within each enterprise is expected to grow aggressively both dimensionally and in complexity, it‘s essential that data storage provisioning be managed equally aggressively. As a matter of fact, Express Computer posts that by taking advantage of things like de-duplication, automated tiering/cataloging, and virtualization, IT architects can reduce storage provisioning time demands by 50 percent and overall costs by 20 percent.
Exactly what is storage provisioning? It‘s the often tedious and complicated process of partitioning/organizing data storage needs using server disk drive space (arranged in order using logical unit numbers or LUNs) in order to improve the efficiency/security/cost effectiveness of a storage area network (SAN). If said data requires interaction between different hosts and are comprised within clusters (as is often the case when data is diversified, stratified and utilized across multiple platforms), this process can become burdensome.
To arrive at a good storage provisioning enterprise deployment plan, consider these best practices:
1. Steps in configuring a storage deploying solution (i.e., NetApp) include:
a.) Input the initial system setup
b.) Set up an Array management configuration
c.) Designate an aggregate; d.) Designate flexible volumes
e.) Designate an additional Ethernet interface & configure Ethernet storage as VIF
f.) Integrate an ISCSI Initiator Group
2. Use thin provisioning instead of overestimating storage capacity; it allows for less-expensive on-demand storage usage strategies.
3. Use de-duplication to eliminate/reduce duplicate data within systems.
4. Use “snapshots” (a frozen-in-the-system, read-only view of an aggregate or volume) to reduce unexpected file backup, directory hierarchies, LUNs and application data.
5. In order to decrease disruption of desktop clients by failure/congestion, select storage systems that support redundant data copies, multiple paths, rapid load reallocation if a storage component/path becomes available, and dynamic load balancing of I/O across copies and paths.
6. When configuring storage devices it‘s best if you configure storage I/O targets as RAID LUNs over multiple spindles; this helps ensure continued accessibility during spindle failure.
7. The best RAID level for a target LUN is dependent on the I/O volume“”for example, how much said volume is utilized to write operations.
8. Higher drive rotational speed ameliorates performance but also up costs; in order to meet I/O VDI workload demands with good response times, place frequently accessed data on spindles with high rotational speeds.
9. Provide the most accessed devices on flash drives to improve system performance.
10. Best storage systems should be able to quickly reallocate a spare spindle and integrate it back into the RAIDS LUN minus the downtime or excessive overhead.
11. Distribute across multiple smaller LUNs to reduce I/O request queuing to improve performance.
12. To avoid misalignment between storage device layout and disk structures (which worsens I/O response tmes), using an EMC storage array, align the data partitions on a 64kb boundary; better yet, align both the VMFS file system and accompanying virtual disks on 64kb track boundaries.
13. Store user files separately from OS desktop in order to make backing up user files and roaming profiles easier.
14. Utilizing a VMware infrastructure can be less expensive than alternatives, allowing different departments to securely share servers and networks.
15. Using a Cisco WAAS in a VDI environment can provide an RDP performance assuring a PC-like experience over the WAN, more VMware view desktops supportable by the WAN, accessibility to local MS Windows print services to different branch users (thus reducing bandwidth demands) and improved VMware view desktop images backups between disaster recovery centers.
16. Utilize real-time compression technology; instead of just compressing only low-activity or infrequently accessed data, it compresses active data by about 80 percent.
17. Utilize context-aware computing in order to ameliorate end-user/system interactions.
18. Become familiar with enterprise application deployment descriptor elements; they will help you understand or set up the configuration of your own storage provisioning deployment strategy.
Storage provisioning will become an increasingly essential part of running servers as datacenters grow in size and number to accommodate our thirst for information. Keeping these tips in mind will give you the insight needed to confront the inevitable problems of increased deployment of enterprise applications.
