Обзор Trellis от Emerson Network Power: оригинальная (английская) версия
Опубликовано: 27.02.2012. AboutDC.ru
В декабре 2011 года компания Emerson Network Power объявила о выпуске Avocent Universal Management Gateway для платформы Trellis. На сегодняшний день о Trellis в России известно немного, поэтому мы решили получить информацию из первых рук - на наши вопросы отвечает Вольфганг Горецки, продакт-менеджер Emerson Network Power по региону EMEA (Wolfgang Goretzki, Product Marketing Manager EMEA):
AboutDC.ru: How long ago did Emerson conceive DCIM development? What was the definition of this concept, what were the objectives of DCIM implementation in DCs?
In the past, Emerson had extensive power and cooling systems related facilities monitoring and management capabilities with a large base of successful customer implementations supported by a large monitoring services business. Based on this experience, Emerson identified the need for more comprehensive data center infrastructure management capabilities for the data center that spanned both facilities and IT systems to address critical customer needs. Emerson has been planning a DCIM solution for several years now and the execution of this strategy started with the acquisition of Aperture in 2008 and Avocent at the end of 2009, in order to gain core competencies in IT infrastructure management.
There are many vendors in the DCIM market and there are many different views to what this might be: some IT vendors provide tools to help track the assets (in other words: datacenter documentation) and plan for changes in the data center (possibly including the capability of capacity planning). Other vendors have power metering solutions to monitor the power consumption of mainly the IT assets, using software combined with intelligent rack PDU´s and inline meters. There are also solutions combining both to have real world power data for accurate capacity planning or even real-time power information and so being able to calculate KPI´s such as the PUE factor. Some tools may include change management and allow to define and execute a process work flow to ensure all steps of the changes being executed. IT infrastructure management solutions are mentioned as well. These provide remote access to the IT assets by utilizing their service processors, KVM-connections or console ports through appliances and ideally offering these access possibilities through one single console. Or tools for monitoring and remote access of the physical infrastructure (the facilities) - these manage the physical infrastructure of the datacenter.
WE define DCIM to comprise ALL of these in ONE holistic solution for both IT- as well as FACILITY- infrastructure management.
Our objective when implementing that DCIM solution was not to use all of those components we already have and somehow interconnect them to a complete solution, but to design that solution completely new: on a new platform with the ability of perfect interaction in between all functions, with a single database describing all the configuration and operational status of the datacenter infrastructure AND the guarantee of true real-time capability.
AboutDC.ru: What do Trellis and Avocent Universal Management Gateway (UMG) look like?
Trellis is a real-time infrastructure optimization platform that enables the unified management of the data center’s IT and facilities infrastructure, allowing for unprecedented insight and action to achieve optimal data center performance. The Trellis platform is comprised of hardware and software and gathers vast amounts of real-time information from IT and facility devices and consolidates it into a simple, meaningful view, enabling the data center to operate at the lowest cost possible and still maintain high levels of availability. The Avocent Universal Management Gateway (UMG) is a multi-purpose appliance that offers consolidated access to facility and IT equipment, making it possible for data centers to execute a unified approach to infrastructure management resulting in greatly reduced cost and more efficient management and control. The UMG is the core component of the DCIM framework. The UMG is both the data collection engine for the Trellis platform applications as well as the closed loop control. Today, the closed loop control is supported through DSView Software but in the future will be available in the Trellis Platform. The UMG combines the functions of KVM over IP, serial console over IP, rack PDU, SP, and environmental management in a single chassis. When used in combination with DSView Software, the solution provides remote access and control to any device that can be controlled by any of the consolidated functions.
AboutDC.ru: What are the building components of Trellis and UMG?
Trellis is a platform which is based on software and hardware:
The Trellis software is modular and can provide all DCIM functions – for managing both the IT- as well as the facility- infrastructure. As these all rely on a common database, the application can be context driven and understand the correlations in between each other, including those between the dynamic IT and the impact to and requirements from the facilities.
The hardware – the Avocent Universal Management Gateway (UMG) plays an important role in the Trellis DCIM solution: it handles all the different protocols and allows to communicate with facilities devices, such as cooling equipment and UPS´s, environmental sensors as well as the IT devices through their service processors, KVM connections or console ports. It allows to monitor, access and control these devices in real-time and therefore to build a closed loop control system for datacenter infrastructure management.
Datacenter managers will benefit from that with a real-time optimized datacenter.
AboutDC.ru: What does the product offer to help optimize DC monitoring and management? (What engineering and IT systems are involved, how data is gathered etc.)
On a single, unified, dynamic platform, the customer will be able to monitor and control the critical infrastructure (IT and facility equipment), manage capacity, track inventory, plan changes, visualize configurations, analyze and calculate energy usage, optimize cooling, power equipment and account for the impact of virtualization on the equipment supporting it. This single platform allows the user to manage the system as a whole, instead of in parts, resulting in higher utilization of existing infrastructure, increased personnel efficiency and lower operating costs. It will collect, aggregate and analyze data points in seconds by connecting natively to the device, versus integrating with other monitoring solutions, to gather historical data or polling equipment in minutes, hours or days only.
This offers the unique benefits to align IT and facilities and extend IT service management (ITSM) principles into the critical infrastructure and drive performance throughout the business.
AboutDC.ru: What are minimum DC size requirements for implementing this DCIM? What are scalability options as DC grows?
Trellis is primarily aiming at datacenters of large enterprises, typically 5,000 square feet or more – that target audience has the strongest desire to achieve higher operational efficiency and especially a better utilization of expensive stranded capacities by managing their datacenter through a holistic solution. However we are planning for a version for smaller businesses as well through a Software as a Service offering which will become available in the first half of 2013.
Now,scaling up would not be an issue at all. Most solutions would likely have quite a challenge to maintain real-time capability collecting and checking large number of data in larger environments. Due to the architecture of the Trellis DCIM solution, which utilizes an appliance for these tasks (the UMG),Trellis can support any size of datacenters: it´s just a matter of applying an appropriate number of appliances.
AboutDC.ru: Is it possible to monitor third-party equipment?
Trellis is intended to be an open system - Trellis will be utilizing standard interfaces (SNMP, BACNET, MODBUS,…), which allow integration of third party devices into the Trellis solution. In addition our professional services team will be available to support integrations.
AboutDC.ru: To what extent is the product modular? What modules are available?
The Trellis platform solves a number of DCIM issues and is designed to allow the user to easily adopt new capabilities as their needs arise. They can invest only in what is needed with the confidence that the technology will be current and applications will be easy to add.
Trellis will consist of the following modules: the Inventory Manager (this is the minimum to start with) – it tracks inventory down to every asset’s floor or rack position and its dependencies. The Change Planner – it allows the datacenter team to track and plan changes. The Site Manager – it tracks and reports on the health of all facility-critical infrastructure devices. The Energy Insight – Its a dashboard that provides datacenter managers full visibility into energy consumption. The Power System Manager – which enables control of the entire datacenter power system and provides a dashboard showing system capacity, utilization and stranded capacity. The Virtual Insight – this gives visibility to both physical and virtual inventory and shows physical capacity needed vs. what is available. And finally, the Cooling System Manager – he offers comprehensive view of cooling system resource utilization and capacity.
AboutDC.ru: What hardware (servers, crosses…) is required?
We are currently benchmarking the various software modules with simulations of varying load and data volumes. This is soon to be completed and we will then have a clear picture about minimum and suggested hardware and software requirements.
AboutDC.ru: What are the application's graphics features? (DC, utility modeling, technology diagrams…) Are there any images?
Visualization of the datacenter static configuration as well as dynamic operational state is key to the Trellis DCIM solution.
Here are some examples – you can look at floorplans incl. consumed capacities , rackviews including connections, visualize the virtualization infrastructure, plan changes graphically, define/view change process work flow charts, look at dashboards to see the current status and understand the impact of changes across the datacenter, view facility layout and asset utilization, get a dynamic view of the power distribution from generator to the rack including visualization of alerts – or view the design and operating state of the cooling system and all of its devices.
AboutDC.ru: Should DC be controlled by a single manager or responsibilities can be shared between several managers?
The main question is, if and how much IT managers and facility managers collaborate today. In many enterprises these departments still work almost independently from each other with facilities having just basic knowledge about the capacities, required by the IT. And the multi-tennant architecture of Trellis certainly can match these situations and provide relevant information to all corresponding users.
However: due to virtualization and private cloud technologies IT has become very dynamic and that increased the risk of unanticipated hotspots or capacity overload. To avoid downtime by that, datacenter managers have increased their buffer capacities of the expensive facilities infrastructure. Trellis, a solution, which understands the correlation between dynamic IT and facilities and which provides a holistic and true real-time view, will significantly reduce expenses and cultivate a stronger collaboration of facilities and IT teams.
AboutDC.ru: What kinds of alarm notification are used? What are DC management algorithms preset (power optimization, video surveillance, placing IT hardware into racks, etc.)? Can you give any examples?
It’s hard to describe in a short paragraph, however let me give you some examples:
- The Trellis platform element libraries are responsible for defining the communication protocols, data points to be monitored and sending control commands to endpoint devices.
- Also contained within the element library are the data points that can be collected from each device and the ability to configure the data points such that an administrator can choose different data points from the delivered set of defaults. For devices that possess the capability, an element library can define control commands that can be issued from the interface, such as cycling the power on a CRAC unit or an outlet on a PDU
- Trellis will have complex event processing capabilities. This is a set of techniques and tools to help the user understand and control event-driven information systems. In the data center critical infrastructure, events do not occur in a vacuum. With complex event processing, the Trellis platform can predict events before they occur and automate routine processes, just like our autonomic nervous system.
- Trellis will gather and aggregate millions of data points in real time and apply rules to discern previously undetected patterns and trends, then presents this information in context. This way the user can immediately grasp the current health of the infrastructure, what impact proposed changes will have on performance and how to fine tune components/zones to maximize efficiency.
- Trellis includes all relevant physical parameters of IT devices and support to place IT equipment in the most efficient way to avoid unbalanced heat dissipation or capacity overload