How it Works
Hp
Loading...
ISS Technology
Volume 7, Number 4
12 pgs201.6 Kb0
Volume 7, Number 5
17 pgs1.51 Mb0
Volume 6, Number 7
20 pgs1.18 Mb0
Table of contents
Loading...

Hp ISS Technology Volume 7, Number 5

ISS Technology Update
Volume 7, Number 5
Keeping you informed of the latest ISS technology
When to use Brocade SAN switches or Virtual Connect-Fibre Channel modules
The latest HP c-Class BladeSystem supports multiple interconnect options based on N_Port ID Virtualization (NPIV) technology, two of which are the HP 4Gb Virtual Connect Fibre Channel module and the HP StorageWorks Brocade 4Gb SAN Switch using Access Gateway Mode. NPIV is an industry-standard Fibre Channel (FC) protocol that makes a single FC port appear as multiple virtual ports, each having its own N_Port ID and virtual world wide name (WWN). NPIV enables several solutions such as HBA and server virtualization and gateway connectivity for multiple servers without adding an additional FC domain to an existing FC fabric.

HP Virtual Connect Fibre Channel Modules

Two HP Virtual Connect Fibre Channel (VC-FC) modules are inserted in the lower interconnect bays of the c-Class BladeSystem enclosure (see Figure 1-1). connect to an external FC SAN fabric. The VC-FC modules allow administrators to organize and manage up to 16 servers in the enclosure as a pool of servers in a SAN. The built-in VC Manager software allows administrators to assign a connection profile to each server bay in the enclosure, not to physical servers, so that SAN connections remain constant even if a server is added or replaced. Furthermore, a server profile can be moved from bay to bay in the enclosure.
Each of the HP VC-FC modules can aggregate up to 16 FC HBA ports through the enclosure mid-plane—together the modules provide connectivity to redundant SANs. By default, each of the four uplinks is configured to aggregate up to 4 FC HBAs. Additionally, each uplink can be configured to aggregate up to 8 or 16 FC HBAs. In essence, each VC-FC Module acts as a pass-through device to the network. Any change to a server is transparent to its associated network. This separates the servers from the SAN, relieving SAN administrators from server maintenance.
1
When installing the HP VC-FC Module in the c-Class enclosure, the HP 1/10Gb VC-Enet Module is also required because it contains the
processor that runs the VC Manager firmware.
1
The VC-FC modules connect to the enclosure mid-plane and feature four uplinks, or N_Ports, which
ISS Technology Update Volume 7, Number 5
Figure 1-1. Port mapping for each VC-FC module installed in the ProLiant c7000 BladeSystem enclosure
The HP VC-FC module provides the following benefits:
• Server management is separated from LAN and SAN management. The SAN needs to be managed at the core switch only,
separate from BladeSystem enclosure management.
• Changing servers takes minutes instead of days, and servers can be added, replaced, and recovered without affecting SANs
or LANs.
• It is interoperable with multiple vendor fabrics (Brocade, McDATA, and Cisco).
• It does not consume a domain, thereby avoiding switch count scalability limit issues in a fabric and eliminating an additional
hop between switches.
• It reduces the number of cables and small form factor plug-ins (SFPs).

HP StorageWorks Brocade 4Gb SAN Switch using Access Gateway Mode

Access Gateway (AG) Mode is a software feature for p-Class and c-Class enclosures that can be enabled in Brocade embedded switches. This feature is introduced in Brocade Fabric Operating System (FOS) v5.2.1b, and it does not require the purchase of additional hardware or software beyond the Brocade-embedded switches themselves. After the AG Mode is software-enabled, these switches act as simple port aggregators using NPIV technology. They use N_Port protocol to connect to any vendor’s NPIV-compliant FC switch, and provide similar features as the VC-FC Module, except for the unique VC support and manageability provided by the HP VC Manager software. Figure 1-2 shows the Brocade 4Gb SAN Switches for the c­Class BladeSystem for a better comparison with the HP VC-FC Module, which works only in the c-Class enclosure.
The Brocade 4Gb SAN Switch for the c-Class enclosure in AG Mode offers a maximum of 24 ports, with a maximum of 16 ports that connect to the blade server HBAs and a maximum of eight external ports that can be used as uplink N_Ports. In AG Mode, all eight external ports act as N_Ports supporting NPIV protocol and can connect to any standard FC switch that supports NPIV-compliant F_Ports. AG Mode, like HP VC-FC, does not consume a domain and, therefore, helps to avoid exceeding domain count limits in large fabrics with multiple blade enclosures.
2
ISS Technology Update Volume 7, Number 5
Figure 1-2. Brocade 4GB SAN Switches installed in a ProLiant c7000 BladeSystem enclosure
The main benefits of the Brocade 4Gb SAN Switch using AG Mode are as follows:
• It is the only NPIV solution for HP p-Class enclosures.
• It provides a 2:1 server-to-uplink ratio for a fully populated enclosure with 16 servers.
• It has a flexible licensing option (12 or 24 ports, with a 12-port upgrade option for the 12-port model)
• It can be repurposed as either Switch Mode or AG Mode, but not in both modes simultaneously.
• It is interoperable with multiple vendor fabrics (Brocade, McDATA, and Cisco).
• It does not consume a domain, thereby avoiding switch count scalability limit issues in a fabric and eliminating an additional
hop between switches.
• It reduces the number of cables and small form factor plug-ins (SFPs).
• It eliminates the need for SAN management inside the enclosure after the initial connections have been configured.

Conclusion

HP VC-FC Module solutions are best suited in customer environments where the ability to manage servers without impacting SANs is a high priority. The Brocade 4Gb SAN Switch using AG Mode may be a better choice for customers with different requirements:
•need more than four FC uplinks.*
• use multiple WWNs on each HBA port (NPIV) on a server running ESX 3.5.*
• are not ready to adopt VC-Enet modules.
• have standardized on a non-HP Virtual Connect LAN environment.
• need Brocade proprietary subtle performance features such as multiple levels of priority.
* This will not be an issue for future HP VC-FC models
3
ISS Technology Update Volume 7, Number 5

Additional resources

For additional information on the topics discussed in this article, visit:
Resource URL
Why Choose HP Virtual Connect Enterprise Manager?
NPIV and Fibre Channel interconnect options for HP BladeSystem c Class white paper
What is NPIV? Web page
http://h20000.www2.hp.com/bc/docs/support/SupportMa nual/c01373909/c01373909.pdf
http://h71028.www7.hp.com/ERC/downloads/4AA1­2234ENW.pdf
http://h71028.www7.hp.com/eNewsletter/cache/584543­0-0-224-121.html

Delivering energy-efficient solutions for the IT power and cooling chain

According to DOE power consumption estimates, U.S. data centers consumed 61 billion kilowatt-hours in 2006. This represents only 1.5 percent of the total U.S. power consumption. However, this power use is nearly doubling every year due the increasing number of new and planned data centers.
Impending environmental regulation is creating a sense of urgency among businesses to improve overall efficiency of existing and planned data centers before legislative action is taken to curb energy consumption.
One of the greatest challenges for HP is also one of its greatest opportunities—helping customers adopt next-generation, high­density platforms in aging and/or crowded data centers while trying to improve the overall energy efficiency of these facilities. Businesses are beginning to buy energy-efficient server and storage platforms; however, the best way for them to improve overall data center efficiency is to change their day-to-day operating behavior. For example, a few years ago, the paradigm shift from centralized computing to distributed computing infrastructures benefited businesses as well as the entire server industry. However, this shift drove a great deal of processor “underutilization” resulting in considerable energy inefficiency.
Today, with average processor utilization hovering around 15 percent, virtualization is the most effective strategy to reduce data center energy use. It can save up to 70 percent of the energy required to power servers. Storage consolidation—spindle reduction—is also key to reducing energy use. If businesses change their operating behaviors to use these strategies, they can achieve their business goals with fewer machines while improving their facilitie’s overall energy efficiency. These behavioral changes (Figure 2-1) are among a list of other steps that can be taken from the “chip to the chiller” to reduce energy use.
4
ISS Technology Update Volume 7, Number 5
Figure 2-1. Behavioral changes
Businesses can take some immediate steps to reduce energy use:
• Low-power processors – The somewhat higher cost of low-power processors has discouraged some customers from requesting
them in their servers. However, customers who understand that the operating cost over the lifetime of a server is much higher than its acquisition cost tend to use low-power processors to improve overall return on investment.
• Energy-efficient power supplies – Recently, there has been dramatic improvement in power supply efficiency. Only 3 years
ago, power supply efficiency was between 65 to 75 percent efficient, depending on the load. Today, HP power supplies can be more than 90% efficient at 50 percent load.
• Power distribution racks – A power distribution rack is a rack-mounted breaker panel that does not use a transformer to step
down the voltage—it’s 208VAC in and 208VAC out. In contrast, transformer-based power distribution devices that step down the voltage from 480VAC to 208VAC have energy losses. Changing to non-transformer-based power distribution can reduce energy losses around 5 to 7 percent.
More importantly, businesses need to understand the HP holistic strategy. This strategy goes beyond the technologies inside server or storage platforms. It involves factors that are centric to the room, for example, the air. First, air is a fluid, which means that we can characterize its behavior in the room using computational fluid dynamics (CFD) modeling. Then, CFD enables us to find ways to control the air to get the most cooling from it. By controlling the air, businesses can begin to adopt next-generation, high-density platforms to get more equipment in the room. This may not reduce energy use; however, it can reduce the rate of increase in energy use as customers adopt new platforms.
In summary, if businesses holistically change their operating behaviors to reduce energy use—or reduce the rate of increase in energy use—they can extend the life of their facilities. These behaviors include virtualization, consolidation, and automation. Factoring in the lead time and cost required to build a new data center, businesses can save millions of dollars for every year they extend the life of their existing facility.
5
ISS Technology Update Volume 7, Number 5

Meet the Expert—Michael Chan

Michael Chan’s 18-year career with Compaq/HP is proof that “timing is everything.” Michael has excelled in signal integrity (SI) engineering due to his discipline and attention to detail. SI, which addresses signal timing and quality, is a critical technology that impacts all levels of server platform design. SI analysis is performed using computer simulation because signaling problems are extremely difficult to diagnose and solve after a system is built. If signaling problems are not understood and solved prior to PCB fabrication, they can cause delays that increase the project cycle and development cost.
Michael began his career creating Spice models and working on Simultaneous Switching Noise Analysis for ASIC package designs. Then he worked on electromagnetic interference (EMI) design for various desktops, portables, servers, array controller cards, and printers. After that he worked on device and system level SI in the portable division. Since then, he has made significant contributions to ISS platform designs, including the first Itanium servers, AMD servers, and now BladeSystem servers. According to Michael’s manager, Mike Pelonero, “The SI aspect of these designs has become more critical as processor, memory, and overall system frequencies have increased as have design densities, especially with regard to blade servers”
Michael’s expertise seems to spill over into some of his hobbies, and vice versa. He enjoys audio amplifier construction, cassette deck repairs, plastic model construction, and stamp collecting. He is the father of three—twins (a girl and boy) age 11 and a boy age 8.
Below are excerpts from an interview with Michael Chan.

Why did you decide to become an engineer?

Ever since I was a kid, I’ve liked to build and repair stuff. In order to build things, one needs to know how to meet the design specifications. At the same time, repairing something requires one to understand how it works and why it broke. It seems like engineering was my destiny.

What has been one of your most interesting research projects?

One of my most interesting research projects was analyzing the effect of high­frequency decoupling capacitors on laptop computers. We used to put a lot of high-frequency decoupling capacitors with many different capacitance values into
Name: Michael Chan Title: Hardware Design Engineer (ISSG) Years at HP: 18 Patents:
• Chan, Michael Y., Leigh, Kevin B.
“Apparatus For Controlling The Impedance of High-Speed Signals on A Printed Circuit Board.” U. S Patent 5764489. June 9, 1998.
• Chan, Michael Y., Leigh, Kevin B.
“Apparatus For Controlling The Impedance of High-Speed Signals on A Printed Circuit Board.” U. S Patent 5986893. November 11,
1999.
• Chan, Michael Y., Riley, Dwight D.
“Point-To-Point Electrical Loading For a Multi-Drop Bus.” U. S. Patent
7099966. August 29, 2006.
University/Degree
• University of Arizona at Tucson:
Bachelor of Science Geosciences
• University of Arizona at Tucson:
Master of Science Electrical Engineering”
our laptop products. The decision to use that many capacitors was based on rules of thumb. This practice increased both the cost of the product as well as component inventory costs.
We made some emission measurements on one laptop and found out that there was very little difference in terms of emission level detected between a laptop that is fully populated with high-frequency decoupling capacitors and one with no capacitors. At that moment, we still did not know if the capacitors were ineffective or whether something caused those capacitors to become ineffective.
We then performed some computer simulations and found out that the capacitor mounting structure had so much inductance that it rendered all the capacitors ineffective. We modified the design of the mounting structure and repeated the simulation. We also reduced number of capacitors, and the emission level improved significantly. We finally decided to reduce the number of high-frequency capacitors as well as the capacitance values of those being used. This significantly reduced the component cost in our future laptop products.
6
Loading...
+ 11 hidden pages
Buy Points How it Works FAQ Contact Us Questions and Suggestions Privacy Policy Cookie Policy Terms of Use