There's a "great white" inside every SAN: and this man-eater's name is complexity.
Similarly, storage administrators set out with good intentions and plans to handle their growing storage needs. By migrating direct-attached storage to storage networks. IT professionals are just now discovering the monster lurking under the surface of their SANs.
The "great white beast" in storage is the growing number of access paths between applications and their storage, and growing complexity between the paths' interdependencies and relationships. This has storage administrators symbolically yearning for a bigger boat; in this case, software that can reliably and quickly change and grow storage networks.
Like the characters in Jaws, IT staffers find themselves poorly equipped to confront the beast, armed with only manual methods and archaic spreadsheets. Adding staff is not the answer, since analyzing the vast number of logical and physical access paths winding through the SAN surpasses the capacity of any number of highly skilled administrators.
Without the management framework to automate and accelerate the change process, the enterprise remains at risk of downtimes, brownouts, security breaches, and loss of customer confidence. A recent Computerworld survey found managing the complexity of storage networks one of IT's top challenges. Even a relatively small SAN can have tens of thousands of potential configuration states. A simple mistake in cabling can become a crippling problem that can compromise security, cause lost data, and waste hours of productivity trying to fix the problem. Worse yet, no easy way exists to find the root cause of such crippling problems--violations that stray from defined business policies.
The Figure highlights what many deem a "complexity crisis" poised to end the growth of SANs. Potential access paths are shown to grow exponentially as the enterprise increases from 60 hosts to 150 hosts--an average sampling of the amount of hosts found at a typical Fortune 1000 enterprise.
Many storage administrators report that SANs run well once they are installed, but fail to adapt well to changes--adding a server or a switch for example. Twenty-five to 35% of SAN changes have at least one error. These errors can create critical problems that may go unnoticed until some later event exposes it, causing catastrophic system failure, security lapses, or loss of business data. As enterprises come to depend on SANs as a reliable service delivered by the IT organization, the business is quickly losing confidence in the SAN's ability to serve as an always-available utility. Like the small-town sheriff, IT needs a bigger boat to solve the problem.
Today's SAN management tools fail to achieve one of the enterprises' top priorities: end-to-end availability and assurance. For a business to realize the full value and ROI of a SAN, the enterprise must be able to make changes accurately and quickly to keep pace with business requirements. Provisioning, monitoring, utilization, and other software tools provide "nice-to-have" resources, but they offer little value if the SAN is unavailable or unstable. Some software tools manage a particular device within the SAN, but cannot account for the myriad of SAN devices and access paths.
All fail to take into account the mission-critical need for flexibility and managing changes and complexity within the SAN. According to Marc Staimer, president of Dragon Slayer Consulting, "Current tools provide administrators with the illusion of control."
The Hidden Menace
Today's SAN is burdened by bottlenecks. Planned and emergency changes take days and weeks to complete, and then cannot be validated to ensure that they were made properly. SAN change cycles average 10-12 days, with an acceleration of 4 days for emergency changes. Beyond the lack of software to troubleshoot and fix errors before, during and after changes, storage administrators need to better control change groups, often dispersed across an organization, and ensure that changes have been made and made properly. Control becomes particularly challenging when change directives must be performed in a precise sequence among disparate groups including operations, storage, and cabling.
Ironically, the very investment a company made in improving storage efficiencies has become an operational logjam. Like the shark rising from the briny deep, the challenges of change management have risen to the surface. They now threaten productivity and business continuity.
However, a solution may be at hand: software that detects fatal errors before, during, and after SAN changes. This technology continuously maps, simulates, and analyzes the entire SAN in order to discover root causes of problems. Predictive change management may just be the bigger boat that storage administrators need for root-cause analysis of SAN problems, and in proactively managing SAN change and growth. Predictive SAN change management will reduce operational complexity, costs, and risk and improve SAN availability, assurance, and customer confidence.
According to analyst firm Gartner, "Improving IT change management processes is generally considered one of the best investments an enterprise can make. Companies that don't properly manage IT changes lose time, money, and efficiency and are subjecting the entire business to undo risk."
Here is a breakdown of how the process could be better addressed through predictive change management software:
SAN change validation: Within the predictive change management software console, the administrator can view the open access paths in the network and see any changes to the SAN along with their analyzed impact on the access path availability, performance and security.
SAN change troubleshooting: Whenever a problem is discovered during the validation phase, its root-cause can be analyzed instantly. This enables the appropriate fix to take place quickly with very limited impact on service level, even before the storage user is aware of any problems.
SAN change audit: The management software also captures a comprehensive change history of all processes and events in order to generate management summary and trending reports, troubleshoot and validate change implementation, and to facilitate the documentation and audit capabilities of all change history and processes.
Planning: With traditional SAN management tools, the planning process remains manually driven, comprised of spreadsheets and other diagrams that do not provide storage administrators visibility into their SAN. By employing predictive change management software, the storage administrator quickly captures and details all required change tasks and actions.
Delegating: The software then creates a simulation of the planned changes, and their projected impact on the network. Change tickets are then created that can be delegated to the respective group or department (e.g., cabling, storage, etc.) for implementation.
Tracking: The software assists in delegation, and coordination of the activities of departments assigned to implement change tasks. The software also logs and tracks every configuration change in the SAN and validates that change tasks have been made correctly, in the proper order and manner.
The benefits of predictive change management become readily apparent when compared with previous methods, improving accuracy, improving operational efficiency and accelerating change times.
As demands for storage capacities rise (the Meta Group projects storage capacities in enterprise data centers increasing 40-60% annually), the technical complexity of storage environments will continue to escalate and the confidence of IT professionals in them will likewise deteriorate.
By evaluating the need for management frameworks to incorporate flexibility and change management models into their functionality, companies will see the next generation in management software come of age. The smartest companies will take advantage of this opportunity to tame their SAN complexity beast, and not become a competitor's shark bait.
Number of potential access paths from hosts to LUNs in core-edge Host Access Paths to LUN 40 277,333 50 906,667 60 2,016,000 90 9,504,000 120 25,920,000 150 54,720,000 Assumes: 10 LUNs per host 30 hosts per 64 switch (2:1 redundancy) 15 hosts per 32 switch (2:1 redundancy) Note: Table made from line graph.
Assaf Levy is vice president of product management and cofounder of Onaro, Inc. (Boston, MA)
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||SAN Trends; storage area network|
|Publication:||Computer Technology Review|
|Date:||Oct 1, 2004|
|Previous Article:||Information security demands a layered approach in IP SANs.|
|Next Article:||RAID-on-a-chip [ROC] processors change the storage equation: Part 2.|