The newsletter for hp certified Professionals



Yüklə 0,53 Mb.
tarix08.08.2018
ölçüsü0,53 Mb.
#61583



January 2003

Volume 2, number 1



Industry Standard Servers

HP ISS technology update





the newsletter for HP Certified Professionals


in this issue

  • ISS engineering problem communication team delivers satisfaction through problem communication 1

  • about this newsletter 1

  • questions and
    answers 1

  • ProLiant BL p-class interconnect switch 2

  • ProLiant performance analyzer 3

  • world class technology deliverables 4

  • recently published papers 5

about this newsletter

ISS recognizes that HP Certified Professionals are advocates who significantly contribute to the sales and support of HP products.

This newsletter strives to be information you can use in your daily tasks. It continues to be a work in progress as we hone the content to meet your needs.

To do that, we need your feedback about what information will be most helpful to you. Please submit your comments to Editor at TechCom@hp.com.


ISS engineering problem communication team delivers satisfaction through problem communication

This newsletter has contained a series of articles on the ISS Engineering Product Support and Quality Team. This series concludes with this article about the Engineering Problem Communication (EPC) Team. EPC minimizes customer and warranty issues for ProLiant products by researching and communicating the cause of problems, the products and operating systems affected, and the availability of the software and spares to implement the fix. And they communicate to:

  • the HP field organizations

  • technical support organizations

  • ProLiant server customers

Because of the outward-facing communication that the team does, you have probably benefited from their work even if you didn’t recognize the source.

research, communicate, and deliver

EPC communicates ProLiant issues that are discovered by customers and escalated to HP Engineering. About 70 percent of the issues are found during development and internal QA, and many are documented before the product ever reaches the customer.

EPC communications also cover topics ranging from best practices and engineering recommendations to Bulletins, which address more serious issues such as hardware damage and data loss. Once the issues are understood and a resolution has been determined (and all aspects checked), the actual communication is written, reviewed, revised, and released.

continued on next page

questions and answers

1. Regarding Microsoft .NET: Why even have four nodes? How are we going to convince customers that they need 8 nodes?

Response:

The biggest advantage to supporting more than 2 nodes is to get better efficiency in utilizing hardware, particularly in an “n+1” configuration. Many customers wish to have the same level of performance after a failover, and a 2-node cluster requires them to configure them in an active/passive configuration. In an n+1 configuration, one server can act as a standby for up to 7 active servers. However, 8-node clustering will most likely not be the driving force for .NET migration. Customers will decide to migrate to .NET for other reasons, and 8-node clustering support will be an added incentive, but probably not the primary driving force.

2. The customer wants to migrate from NT4 to either Windows 2000 or to .NET. They don’t know which one they should explore. Any thoughts?

Response:

Much of the design work will be the same for either scenario, so it should be started as soon as possible even if they elect to wait for .NET. In fact, if they haven’t already started, then they are unlikely to be ready before June 2003 assuming they are anything other than a small organization.

As a rule, it is sensible not to deploy significant infrastructure on a product until that product has had a chance to ‘stabilize.’ In the .NET case, shipping SP1 would be the indication of stabilization.

.NET is scheduled to ship early in 2003. Applying the rule above, .NET should expect widespread deployment sometime after June 2003. If your customer wants to do limited pilots and other design checkout work, then .NET is an acceptable choice. If they want to do the complete migration before June 2003, then it should be Windows 2000. Then they will have to upgrade to .NET later.

The message is not that .NET is unstable, as quite the opposite is the case for RC1. The customer simply needs to understand the expected timetable.



continued on page 4




ProLiant BL p-class interconnect switch

Since many server blades are packaged in a small space, the number of network cables within this space can quickly become overwhelming. The ProLiant BL p-Class GbE Interconnect Switch is designed to dramatically reduce the number of network cables attached to the rear of the ProLiant BL p-Class server blade enclosure.

A pair of redundant hot-plug and hot-remove GbE Interconnect Switches mount as bookends from the front of the server enclosure (as shown above). The fully redundant hot pluggable design supports industry standard, enterprise-class network features. The switches are configurable with the included browser-based, command line, and menu-driven interfaces.

A next generation HP ProLiant BL Interconnect Switch (GbE2 Switch) is targeted for availability in mid-2Q03.  The GbE2 switch provides gigabit performance on all switch ports.  The gigabit performance means that gigabit network adapter (NIC) speeds to 1000 Mbps are fully supported on the ProLiant BL20p server with the NC7780 upgrade as well as the new ProLiant BL20p G2 and BL40p servers. The redundant intra-switch crosslinks, front panel Ethernet ports, and all uplinks have all been upgraded to gigabit.  Four gigabit uplinks are provided per switch available as either 10/ 100/ 1000T (C-GbE2 kit) or 1000SX (F-GbE2 kit).

Not only have the switch ports been upgraded to gigabit, but many additional network features are also provided. Examples of the GbE2 network software enhancement include:


  • Multiple spanning tree protocol IEEE 802.1S providing up to 16 spanning tree domains per switch.  The interconnect switch will now consider VLANs when spanning tree blocks redundant paths to avoid network loops.

  • 9K Jumbo frames that improve performance by increasing application throughput and decreasing server CPU utilization.

  • Two firmware images in memory plus a third runtime version enhancing serviceability and availability.

  • Secure shell (SSH), SCP file transfer, and RADIUS client user dial-in authentication and authorization all significantly increasing system security.

  • Syslog support for additional network event logging.

Designed with investment protection and longevity in mind, the ProLiant BL p-Class GbE2 Interconnect Switch is fully compatible with existing p-Class server blades, server blade enclosures, and all other p-Class infrastructure components. The GbE2 switch supports any combination of ProLiant BL20p, BL20p G2, and BL40p servers. Customer field upgrades from the GbE to the GbE2 switch will be supported without any hardware upgrade required (other than the switch).  To continue to support the most enterprise and bandwidth intensive applications, the GbE2 switch can be upgraded with future options including:

  • Layer 3 switching firmware upgrade.

  • Layer 47 content delivery switching option. All GbE2 switches support an upcoming mezzanine option card that upgrades the switch from layer 3 to layer 47.

  • 10GbE uplink option–-the existing  gigabit uplinks on each switch may be upgraded to at least a 10 gigabit high bandwidth port each using an upcoming interconnect uplink module option.

continued on next page

ISS engineering problem communication team
(continued from page 1)


The topics sometimes sound negative, but the information is very welcome. When customers are able to fix problems on their own and avoid calling HP, customer satisfaction goes up and service costs go down.

A study conducted in 2002 of subscribers to ActiveUpdate showed a $4 million dollar savings annually when issues are communicated and sent proactively to customers. EPC delivers content to ActiveUpdate and several other proactive delivery tools that are noted below.



for more information

With communication as their mission (and part of their name), the EPC team has many avenues to communicate their deliverables. Examples of communicated topics include:



  • Software fixes

  • Firmware upgrades

  • Known workarounds

Check out the URLs below for more examples of the team’s work:

Channel Services Network (CSN):
http://csn.compaq.com/csn/default.asp

Reference Library:
http://wwss1pro.compaq.com/support/reference_library/viewreferencelibrary.asp

ActiveUpdate:
http://www.compaq.com/products/servers/management/activeupdate/

PCN: (retiring in 2Q 2002)
http://www.compaq.com/support/files/server/us/pcn.html

Subscriber’s Choice:
(replaced PCN in 1Q 2002)
http://www.hp.com/united-states/subscribe

ProLiant performance analyzer

Many customers have concerns about performance status and trending of their servers. They need to be able to respond to performance issues before those concerns become customer response issues. And, to manage their limited and sometimes shrinking budgets, customers need to be able to see performance issues over weeks and months, rather than in increments of minutes and hours.

The ProLiant Performance Analyzer (PPA) provides customers a tool to help address the concerns they have about their servers. PPA detects and analyzes hardware bottlenecks that result in performance bottlenecks on ProLiant servers. The PPA, available as part of the ProLiant Essentials Performance Management Pack (PMP), monitors the performance of key server subsystems on one or more ProLiant servers. Enterprise customers needing to enhance performance or to detect system performance bottlenecks can deploy PPA to:


  • Monitor ProLiant server system performance

  • Monitor many ProLiant servers from a single point and collect the data

  • Proactively deal with performance issues

  • Collate and report performance information for planning purposes

performance monitoring

Customers benefit from performance monitoring within several contexts. Some customers need a tool to help them identify and take advantage of unused or underused resources in their system. In many corporations, 24x7 availability of information technology (IT) assets, often with specific performance requirements, is critical. The proactive performance responses that PPA offers fit very well into those environments.



monitoring configurations

To eliminate concerns about having to commit lots of servers or having to set up monitoring and take it down because of the overhead, customers should understand the configurations for monitoring and how many servers can be monitored in which configurations.

PPA converts a typical enterprise configuration into a monitored environment through one ProLiant server with PPA and Insight Manager installed. This server then monitors the other ProLiant servers. The monitoring server can perform other tasks as well, depending on the level of results logging. If one server monitors the other servers without logging the results, a midrange dual processor ProLiant server can monitor over 100 servers with a refresh rate between 1 and 2 minutes. This means that with 100 servers, a single ProLiant server can check the performance data on each monitored server every 1 to 2 minutes. This configuration provides alerting for proactive responses through Insight Manager.

Configurations that include logging and reporting the data from all monitored ProLiant servers require a database on the monitoring server. PPA supports both MSDE, which is included with Insight Manager 7, and MS SQL Server 2000. The database choice depends on the number of servers to be monitored and logged. A dual processor ProLiant server could log the results from tens of servers using MSDE, and the cost for the set up is fairly low. To log results from more than 50 servers, the monitoring server would need MS SQL Server 2000. With SQL, a dual processor ProLiant DL380 server could monitor hundreds of servers.



futures

As PPA evolves, recommendations to eliminate bottlenecks will be emphasized. These recommendations will build on the detection process but won’t always be a simple answer of “replace the hardware.” The recommendations will be based on system-wide utilization and balancing, not on pushing hardware.

In the immediate future, each new server announcement will contain subordinate releases of PPA that will be available for downloading from the HP website.

for more information

A complete description of the ProLiant Essentials Performance Management Pack, including the ProLiant Performance Analyzer, is available at:



h18004.www1.hp.com/products/servers/proliantessentials/wmpdescription.html

ProLiant BL interconnect switch
(continued from page 2)


SAN connectivity

The ProLiant BL p-Class GbE2 Interconnect Switch also provides pass-through of the ProLiant BL20p G2 fibre channel storage signals. A GbE2 FC SAN connectivity option includes a fibre channel interconnect module that is inserted into the open slot at the rear of each switch, similar to the existing uplink interconnect modules.  Each FC interconnect module (one per switch) will include eight optical transceivers supporting full enclosure of ProLiant BL20 G2 servers, each with the dual FC HBA card.  The GbE switch FC SAN connectivity option availability is targeted for mid-2Q03.  It may be ordered with the GbE2 switch or purchased later and upgraded on-site by the end-user.



for more information

For more information on the E2 Switch and solutions that the ProLiant BL p-Class servers provide, please see the following URL, http://h18004.www1.hp.com/products/servers/proliant-bl/p-class/index.html


questions and answers
(continued from page 1)


3. We have a Linux customer who does not want to use a browser to connect to the iLO board. In the past he used the IRC function on ProLiant servers. This functionality is no longer available in ProLiant servers. All IRC functions are integrated onto the iLO board. The customer wants to remotely connect to the server. To this end, he uses a dial-in server. He wants to connect from this server through a Telnet session to the RIB/iLO board. With IRC it was possible to reboot the server. Can one remotely reset the server through a Telnet session? This customer does not want to use a browser because his field engineers are remotely dialing in and have a 9600 baud connection to the iLO board. Using the browser applet from iLO requires downloading the 6MB java applet. This will take too much time to load. Does anybody know how we can reset the server through a Telnet session with the iLO board?

continued on next page

Response:

At this time, it is not possible to use Telnet to reset the server via iLO.  I suggest using the XML scripting language to remotely reset the server without using a browser.  The XML scripting language is described in the iLO User Guide in Chapter 7, "Group Administration and Scripting," and in Appendix B, "Remote Insight Board Command Language for iLO."  The User Guide may be found on the web site.  Start at http://www.hp.com/servers/lights-out. IRC is serial port-based. Are you looking for a serial port solution or a network-based solution?  The XML scripting mentioned above is network-based, not serial port-based.

Also, it is not necessary to load the Remote Console applet just to use the browser interface to reset power on the server.  The Virtual Power Button does not use an applet, it only uses HTML and JavaScript.  It still may be somewhat slow at 9600 bps, but it is not necessary to sit through a large applet download just to reset power on the server.

The iLO does not support reset through its Telnet interface. We are looking at supporting this feature in the future. The java applet used in iLO is significantly smaller in size than mentioned here.  We do not recommend a dial-up connection of less than 36 Kbps.

I believe the 6 MB "java applet" mentioned is actually the Sun Java Virtual Machine (JVM). It is true that the JVM is approximately 6 MB in size, and it would make one very uncomfortable to load that from a slow connection. However, it needs to be loaded only once, and it could be loaded via CD or whatever other mechanism the field engineers use to install the OS and applications on their machines. Surely they don't load their OS via a 9600 baud connection.

Let's go a step further. The JVM is not required in order to use iLO if they do not want to use the virtual floppy mechanism. To simply browse into iLO and power the server on/off (reset) does not require the JVM to be present on the field engineer's machine. The other dependency on the JVM is the single-cursor remote graphical console. If they choose instead to use the dual-cursor remote graphical console, the JVM is not required. So, the 6 MB JVM should certainly not be the show stopper here.

In reference to the sizes of the web pages and java applets, they are relatively small (in the kilobyte range), so they could be used at 9600 baud. However, we recommend a dial-up bandwidth of 56K (effective bit rates of 36K  48K).

4. The DL580G2 has 4096 MB of memory with 2048 MB as online spare memory. The Windows 2000 boot initialization screen shows 3840 MB memory. System Summary in the Computer Management console also reports 3840 MB of memory. Why the discrepancy?

Response:

If the total memory is 6GB in the system, with 2 GB reserved for online, then the reported information is correct. Some memory is reserved by the OS and systems. I believe the customer misconfigured the online spare. First, the 6GB scenario does not make sense—your online spare bank has to be larger than your standard bank. You cannot spare 4GB with 2GB. If the customer’s intention is to spare 2GB with 2GB (a total of 4GB) then the configuration is not correct. The best way to get to the bottom of this is to check the ROM POST message. It should state how much memory is reserved for the online spare and system memory.

world class technology deliverables

HP Industry Standard Servers Technology Communications Team provides technology deliverables to customers, to HP CP members, and to field systems engineers. These deliverables take many forms including demonstration videos, newsletters, technology briefs and technology profiles, conferences, and product overviews. Recently 12 of these deliverables were entered in the annual Society for Technical Communications Technical Publications Competition. Of the 12 entered, 9 won an award. The list of award-winning deliverables is below. Some of them are available only from the Tech Comm web site, but over half are available on the public site.





questions and answers continued from page 3



Award

Deliverable

URL

Award of Merit

Implementing Microsoft Windows 2000 with Service Pack 3 on Industry-Standard Servers Integration Note

ftp://ftp.compaq.com/pub/supportinformation/papers/172n-0802a-wwen_rev2_us.pdf

Award of Merit

History of Innovation and Value-Add in Compaq Industry Standard Servers Integration Note

ftp://ftp.compaq.com/pub/supportinformation/papers/15vh-0602c-wwen_rev8_us.pdf

Award of Merit

ProLiant DL590/64 Server Technology Brief

ftp://ftp.compaq.com/pub/supportinformation/papers/tc020702tb_rev0_us.pdf

Award of Merit

Compaq ProLiant BL Strategy for Rapid Deployment and Reprovisioning of High-Density Servers Technology Brief

ftp://ftp.compaq.com/pub/supportinformation/papers/tc020402tb_rev0_us.pdf





recently published papers from HP Industry Standard Servers

title

URL

Best Practices for Remote Insight Lights-Out Edition – RILOE and RILOE II

ftp://ftp.compaq.com/pub/supportinformation/papers/tc030103wp_rev0_us.pdf

Clustering Insight Manager 7 on Microsoft Cluster Services

ftp://ftp.compaq.com/pub/products/servers/management/cim7/CLSTCIM7.pdf

Cooling Design: Walking the Path to a Cool ProLiant Server

ftp://ftp.compaq.com/pub/supportinformation/papers/tc030101tb_rev0_us.pdf

Customizing Reports from Insight Manager 7

ftp://ftp.compaq.com/pub/products/servers/management/cim7/Reports.pdf

Fault Management of Storage Works in ProLiant Server Environments

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-fault-mgmt.pdf

Implementing Microsoft Windows .NET Server 2003 Release Candidate 2 on ProLiant Servers

ftp://ftp.compaq.com/pub/partners/microsoft/infolib/osintegration/tc021203in.pdf

Initial Deployment of the ProLiant Support Paq Using Insight Manager 7

ftp://ftp.compaq.com/pub/products/servers/management/cim7/CIM7dply.pdf

Insight Manager 7 Device Type Manager

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-device.pdf

Insight Manager 7 SNMP Extensions

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-snmp.pdf

Integrating HP Intelligent Manageability in Service Delivery: Event Monitoring and Notification

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-service.pdf

Migrating Compaq Insight Manager (WIN32) to Insight Manager 7

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-migrate.pdf

Novell NetWare 6 performance tuning guidelines for ProLiant servers

ftp://ftp.compaq.com/pub/supportinformation/papers/tc021201in_rev0_us.pdf

PCI-X 2.0: PCI Bus Leadership Continues

ftp://ftp.compaq.com/pub/supportinformation/papers/tc021202tp_rev0_us.pdf

Understanding Insight Manager 7 Security

ftp://ftp.compaq.com/pub/products/servers/management/cim7/cim7-security.pdf

Upgrading to Novell NetWare 6 on ProLiant Servers

ftp://ftp.compaq.com/pub/supportinformation/papers/tc021205in_rev1_us.pdf




© 2003 Hewlett-Packard Company. HP, Compaq, the HP logo, ProLiant, Active Answers, and ActiveUpdate are trademarks of the Hewlett-Packard Company in the U.S. and/or other countries. Windows is a registered trademark of Microsoft Corporation. Linux is a registered trademark of Linus Torvalds.

Yüklə 0,53 Mb.

Dostları ilə paylaş:




Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur ©genderi.org 2024
rəhbərliyinə müraciət

    Ana səhifə