Data Center Monitoring, Management and Control: Stacking it All Up
Published on March 2, 2015,
by
Within the data center there are multiple physical layer monitoring systems commonly used to achieve operational efficiency, availability and uptime. While these different systems often overlap in some minimal fashion in their capabilities as well as which devices they are monitoring, each has a specific purpose and focus and is required to derive maximum value. Each of these available systems is well suited for their intended domain, but may not address all of the bigger challenge of managing the data center comprehensively. So while a noble goal has always been to reduce the number of management tools in use, the reality of managing a data center today is that a strategic management plan includes several of the available solutions, each working together in some integrated fashion. And when used in combination, can form the basis of a comprehensive power management strategy, reducing costs and increasing IT service delivery.
These physical layer solutions fit into four broad categories: Building Management, Element Managers, DCIM and SNMP Monitors.
Some background on each of these categories:
Building Management System (BMS): Facilities professionals have a large investment in BMS’ and rely on them for monitoring and controlling their mechanical and electrical facilities gear. In most cases they were deployed at the time when the building structure was originally constructed. BMS systems were authored at a time where the amount of instrumentation was fairly limited, and the amount of control that could be executed was very macro. In most cases, they have fairly simple logic to monitor and take control of the contained facilities-level systems. While their usage with non-facilities devices has been attempted, it is typically a losing battle due to the rigid software structure and limited/costly access to the logic.
Element Managers: These systems support the major active components of the data center and tie very closely to specific manufacturers’ products. They can be thought of as the intelligent extensions of the hardware devices themselves, in many cases exposing capabilities that go beyond those that can be found within each device. Most facilities devices and nearly all IT manufacturers offer element managers for their specific devices. Element managers provide a terrific source of data and yet are usually an untapped opportunity for integration with other systems in the data center.
Data Center Infrastructure Management (DCIM): As a comparatively newer category of solution, “DCIM” usually means different things to different organizations. Depending on the functionality that they are interested in, their definition changes. That said, DCIM solutions can usually be found centered in one of two camps: IT-centric functions (like asset lifecycle management or workflow design and execution) or Facilities-centric (where monitoring of a wide range of energy-related devices is key). While still a new category of solutions just a few years ago, in 2015 this is becoming a common and high priority buy for forward looking Data Center managers and their business analysts.
SNMP Monitors: Great tools for the price (since many of these are open-source) depending on the size of your organization. The old rule, “you get what you pay for” applies here since they often require a lot of maintenance, customization and scripting that you may or may not have the time to do and in fact may not be a great choice for a commercial entity of any scale.
Can’t we all just Getting Along?
So with all of the different systems mentioned above, the water can get pretty muddy pretty quickly. In some cases there is a tendency to try and force one or more of these systems into providing capabilities that are simply out of their natural scope, in other cases there are just so many options to consider that it becomes overwhelming with the end result being that nothing gets implemented. That said, understanding the problem at hand and then crafting a plan to address it is simpler than you think.
Things to consider when looking for a management plan for the data center:
1) Identify your goals in writing. Sit down and determine what you are trying to do, what problems you are trying to solve, and THEN prioritize these needs. Discuss these needs with your peers to gain consensus and assure that their needs are also reflected in ‘the plan’. Before you start down the road of evaluation or implementation of any software based system within your facility make sure you have this written target. Like smelling perfume, it is hard to stay objective unless you have a hard plan to test your efforts against. Without a formal plan, many of the original goals are often forgotten, and priorities becomes fairly chaotic. In many cases, this informal process creates an environment of ineffectiveness.
2) Multiple software systems are not a bad thing if they can be integrated. The key is integration of the data so that a single source of truth has been defined. How and where that information is presented and acted upon is less of an issue and integration doesn’t allows mean forming a literal “single pane of glass”. More important is the “Single Source of Truth” where each component is treated as the source provider for the data most appropriate. Examples of this are: BMS systems provide Chiller status, IPMI systems provide server power supply status, and PDU element managers provide the actual load on their PDU phases or outlets.
3) Don’t forget about the people that have to maintain multiple different hardware products. The people element translates into OPEX costs. Remember that each hardware choice comes with a unique set of software capabilities that require training and support. Limiting the number of hardware and software systems can literally save hundreds of hours in configuration and software/firmware upgrades per month which can translate to 15%-20% of the total costs of ownership.
4) When choosing multiple tools, look for systems that have interfaces, like SNMP and xml based tools like SOAP and REST that can easily integrate with each other. Remember it is the combination of data sources that creates business value. These interfaces must be accessible easily and on a regular basis to create long-term value.
5) Look for manufacturers that have built a reputation on openness and have forged partnerships with the industry’s most chosen complementary products. Work with these mature manufactures that have field-proven results and a desire to discuss YOUR goals, not just their features.
Data center management has been with us for a very long time, and as the number of choices has grown, so has the complexity of doing so effectively. What has changed is the scale of the data center and the urgency to start making very strategic choices, not simply keeping things running the way they did half a dozen years ago. CIOs and their executive leaders are looking for thought leadership, innovation and agility. They are looking for business accountability, not repetition of older processes. And you’ll see that choosing to take NO action is really no longer an option. The most successful IT professionals realize that have a very different set of physical layer challenges than those of half a dozen years ago, and their sense of urgency and innovation is admired and rewarded. How do your management choices stack up?
Within the data center there are multiple physical layer monitoring systems commonly used to achieve operational efficiency, availability and uptime. While these different systems often overlap in some minimal fashion in their capabilities as well as which devices they are monitoring, each has a specific purpose and focus and is required to derive maximum value. Each of these available systems is well suited for their intended domain, but may not address all of the bigger challenge of managing the data center comprehensively. So while a noble goal has always been to reduce the number of management tools in use, the reality of managing a data center today is that a strategic management plan includes several of the available solutions, each working together in some integrated fashion. And when used in combination, can form the basis of a comprehensive power management strategy, reducing costs and increasing IT service delivery.
These physical layer solutions fit into four broad categories: Building Management, Element Managers, DCIM and SNMP Monitors.
Some background on each of these categories:
Building Management System (BMS): Facilities professionals have a large investment in BMS’ and rely on them for monitoring and controlling their mechanical and electrical facilities gear. In most cases they were deployed at the time when the building structure was originally constructed. BMS systems were authored at a time where the amount of instrumentation was fairly limited, and the amount of control that could be executed was very macro. In most cases, they have fairly simple logic to monitor and take control of the contained facilities-level systems. While their usage with non-facilities devices has been attempted, it is typically a losing battle due to the rigid software structure and limited/costly access to the logic.
Element Managers: These systems support the major active components of the data center and tie very closely to specific manufacturers’ products. They can be thought of as the intelligent extensions of the hardware devices themselves, in many cases exposing capabilities that go beyond those that can be found within each device. Most facilities devices and nearly all IT manufacturers offer element managers for their specific devices. Element managers provide a terrific source of data and yet are usually an untapped opportunity for integration with other systems in the data center.
Data Center Infrastructure Management (DCIM): As a comparatively newer category of solution, “DCIM” usually means different things to different organizations. Depending on the functionality that they are interested in, their definition changes. That said, DCIM solutions can usually be found centered in one of two camps: IT-centric functions (like asset lifecycle management or workflow design and execution) or Facilities-centric (where monitoring of a wide range of energy-related devices is key). While still a new category of solutions just a few years ago, in 2015 this is becoming a common and high priority buy for forward looking Data Center managers and their business analysts.
SNMP Monitors: Great tools for the price (since many of these are open-source) depending on the size of your organization. The old rule, “you get what you pay for” applies here since they often require a lot of maintenance, customization and scripting that you may or may not have the time to do and in fact may not be a great choice for a commercial entity of any scale.
Can’t we all just Getting Along?
So with all of the different systems mentioned above, the water can get pretty muddy pretty quickly. In some cases there is a tendency to try and force one or more of these systems into providing capabilities that are simply out of their natural scope, in other cases there are just so many options to consider that it becomes overwhelming with the end result being that nothing gets implemented. That said, understanding the problem at hand and then crafting a plan to address it is simpler than you think.
Things to consider when looking for a management plan for the data center:
1) Identify your goals in writing. Sit down and determine what you are trying to do, what problems you are trying to solve, and THEN prioritize these needs. Discuss these needs with your peers to gain consensus and assure that their needs are also reflected in ‘the plan’. Before you start down the road of evaluation or implementation of any software based system within your facility make sure you have this written target. Like smelling perfume, it is hard to stay objective unless you have a hard plan to test your efforts against. Without a formal plan, many of the original goals are often forgotten, and priorities becomes fairly chaotic. In many cases, this informal process creates an environment of ineffectiveness.
2) Multiple software systems are not a bad thing if they can be integrated. The key is integration of the data so that a single source of truth has been defined. How and where that information is presented and acted upon is less of an issue and integration doesn’t allows mean forming a literal “single pane of glass”. More important is the “Single Source of Truth” where each component is treated as the source provider for the data most appropriate. Examples of this are: BMS systems provide Chiller status, IPMI systems provide server power supply status, and PDU element managers provide the actual load on their PDU phases or outlets.
3) Don’t forget about the people that have to maintain multiple different hardware products. The people element translates into OPEX costs. Remember that each hardware choice comes with a unique set of software capabilities that require training and support. Limiting the number of hardware and software systems can literally save hundreds of hours in configuration and software/firmware upgrades per month which can translate to 15%-20% of the total costs of ownership.
4) When choosing multiple tools, look for systems that have interfaces, like SNMP and xml based tools like SOAP and REST that can easily integrate with each other. Remember it is the combination of data sources that creates business value. These interfaces must be accessible easily and on a regular basis to create long-term value.
5) Look for manufacturers that have built a reputation on openness and have forged partnerships with the industry’s most chosen complementary products. Work with these mature manufactures that have field-proven results and a desire to discuss YOUR goals, not just their features.
Data center management has been with us for a very long time, and as the number of choices has grown, so has the complexity of doing so effectively. What has changed is the scale of the data center and the urgency to start making very strategic choices, not simply keeping things running the way they did half a dozen years ago. CIOs and their executive leaders are looking for thought leadership, innovation and agility. They are looking for business accountability, not repetition of older processes. And you’ll see that choosing to take NO action is really no longer an option. The most successful IT professionals realize that have a very different set of physical layer challenges than those of half a dozen years ago, and their sense of urgency and innovation is admired and rewarded. How do your management choices stack up?
