FOCUS CLOUD INFRASTRUCTURE
Issue 15, April/May
MANAGING THE PHYSICAL LAYER OF CLOUD INFRASTRUCTURE
By Ambrose McNevin
available wherever it is needed. When moving around VMs they don’t expect trouble. Part of the issue is that they no longer move physical servers. When they did there was more visibility to the power issue, like plugging the thing into a rack-mounted PDU,” he says. If the IT manager is moving virtual loads to an area that is already over capacity or that might impact another rack, they should use power management software that is integrated with the virtualization platform.
VIRTUAL HUMAN D
epending on your responsibilities, when moving through the cloud stack, you either start with or eventually arrive at the physical power and cooling infrastructure, where without proper design, deployment and management talk of cloud delivery remains ethereal at best.
Virtualization underpins cloud computing. Virtualization is about elasticity and having capacity where you want it. It is the enabler to move loads around, which in turn needs modularity at the supporting infrastructure in the physical layer. Software is needed there to manage, regulate and adjust power capacity.
Talk to APC by Schneider Electric about cloud and it will tell you about software. Henrik Leerberg, product line director for data center software at APC, accepts that APC is not universally known for its software. “We are still perceived in the market as a hardware vendor... of three-phase UPSs. But we have data center software labs in the US, India, Tokyo and Denmark,” Leerberg says.
For APC, it is about allowing the UPS to dictate where the load resides. “That is the way we are going, but companies are not ready to automate yet. They [data center operators and IT people] are always managing risk – they are cautious about letting a new generation of technology control the environment,” Leerberg says. “The guys moving virtualized loads around consider the physical infrastructure to be
42
www.datacenterdynamics.com
A real virtual machine manager – that is, a human being – understands a scenario such as a failure; for example, when what is needed is to move all those virtual machines to another part of the data center and another UPS. But often they don’t know where the physical servers are and that does matter if you are putting a VM onto a server. Leerberg gives the example of a European fi nance house IT team and describes it as fi ve guys managing 3,000 VMs. “We see many IT departments moving up in abstraction. The IT servers are now part of the data center so it is crucial for them to
SUPPLY
Physical infrastructure Cooling, power, racks
Building management and physical infrastructure management software and alarms
have information from the physical layer.”
The DCIM promise is to be able to tell exactly where those VMs reside on physical servers and to dynamically match the power provision to the load. “Think of the data center as a computer: users want software that controls that. The operating system that runs on it is virtualization, and the software will know what to speed up so the computer runs at its most effi cient,” he says.
SEMI CLOUD
Customers are now asking for right-sized data centers, and in terms of a hardware solution that also means a high degree of utilization, and using software to utilize that capacity and minimize risks. The situation today is that there are semi-dynamic cloud-enabled data centers.
If you want to go further, you need to consider major capex investments and culture change, where the facilities people wrest control of the physical IT assets from the IT department. That will only happen when facilities proves it can provide load-matched, dynamic power and cooling at maximum energy effi ciency. n
DEMAND IT equipment
Servers, routers, switches, virtual jobs
Enterprise and network management software and alarms
Source: APC by Schneider Electric
BEST DATA CENTER INFRASTRUCTURE PRACTICES TO PREPARE FOR CLOUD
1. Increase data center density With the right infrastructure, enterprises can increase the capacity and effi ciency of existing facilities, delaying the need to tap into the cloud, or creating the capacity to support an internal cloud.
2. Increase effi ciency at reduced loads Even with cloud computing, data centers will require some head room to handle normal peaks in demand. Infrastructure systems that can operate effi ciently at reduced loads reduce the cost of designing in that head room.
3. Don’t lose sight of availability Availability is still the highest priority. Ensure eff orts to reduce costs don’t impact availability, or they likely will backfi re.
4. Leverage infrastructure management to get a handle on costs The key to managing the balance between internal and external resources will be ongoing visibility into costs. Infrastructure management visibility.
technologies deliver that
Source: Emerson Network Power Read the full Emerson White Paper online at www.datacenterdynamics. com/whitepapers
UPS PDU
CHILLER A/C
EQUIPMENT
SERVERS SERVERS SERVERS SWITCHES STORAGE STORAGE
NETWORK CABINETS
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64