Tag Archives: HP

XW8400 reloaded

Posted on by
2

Hallo zusammen,

ich habe auch das eine Weile vor mir her geschoben, das Recycling einer XW8400. Die gab es für kleines Geld beim Broker und mit zwei Xeons und 8GB RAM, 19,2GB Speicher-Bandbreite eine doch beachtliche Rennmaschine. Die SAS Platte die mit kam ein wenig klein, DVD Rom ok, aber zu Hause vielleicht nicht das Mittel der Wahl. 6 Port SATA und 4 Port SAS RAID Controller an Bord.

Eine Weile lief die Kiste als ESXi Host und hatte schon eine 2TB Platte von WD spendiert bekommen. Jetzt stand auch die Runderneuerung des PCs meiner Angetrauten an und irgendwie roch das nach dem großen Rundumschlag.

Die Mods dato: eine GTX660 von Asus, Blueray RW Laufwerk und DVD RW Laufwerk – beide von Pioneer. Als Nachbrenner eine Samsung 840er SSD mit 128 GB. Dazu einen internen Icybox Card-Reader mit allem F*** und Feuerstein. Dann Neuinstallation mit Windows 7 Professional.

Alles in allem in den alten Hobel für ca. 450 EUR Teile nachgerüstet und ich muss sagen Hut ab. Respektable 13370 Punkte im 3DMark2006 ohne allzu viel Feintuning und das im Rahmen von Taschengeld-Budget. Windows 7 Performance Index 7,2. Ich hätte bei 7 Jahre altem Blech nicht mit ganz so viel Rums gerechnet. Dabei melden die CPUs zunächst dass sie noch um 1/3 runtergetaktet arbeiten.

Eventuell doch um die 3GHz Quadcore Pumpen bemühen, aber die scheinen noch in Gold aufgewogen zu werden.

Die Erste Runde Mass Effekt 3 mit allem möglichen Grafikschnickschnack aktiviert – den man sich bisher immer gerne verkniffen hat – war ein echtes Erlebnis.

kyp. F.

HP-UX Queue Depth

Posted on by
Reply

Just for the occasion … Setting HBA queue depth length on HP-UX systems with an SCSI I/O control command:

scsictl -a -m ir /dev/rdsk/cXtXdX

delivers the current queue- depth on the questionable device cXtXdX, where the Xes refer to the IDs in your systems.

scsictl -m queue_depth=21 /dev/rdsk/cxtxdx

sets the queue-depth to e.g.21.

The value of your queuedepth depends on the supported values of your storage and are given in a range from 1 to 255. By chance these values may conflict or fit together and well result in dedicated HBAs in dedicated zones.

In my case they fit together by default with HP EVA and HDS VSP, keeping in mind that there are changes needed if the EVA systems have to handle extremely large single files within a single LUN.

DCUCI: Datacenter Unified Computing Implementation

Posted on by
Reply

This week I visited the DCUCI training. The best class I had in a reasonable while. There could have been more labs and the marketing coverage – high and wide introduction was much to much considering the fact that there were two four hour eLearning-sessions upfront which covered this stuff more accurate.

Besides this annoying waste of valuable labtime it was a really great introduction into the Unified Computing System world of CISCO. Although it was not feasible to introduce the UCS preparation and uplink configuration part, covered in an appendix, everything else was covered fair. Introducing the pool concepts, updating templates for ports – interfaces, policies and in the server profile it gave a complete introduction into the hows and whens, even the caveats of the CISCO interpretation of stateless servers.

Even the connection and distributed switching integration into VMWare vSphere with the Nexus 1000v as well as the VM-FEX approach was discussed. The labs cover preferably the Nexus 1000v but due to our smart trainer we implemented the VM-FEX approach as well.

Down this road we found a lot of caveats which never have been covered in the technical deep dive classes offered from CISCO for end customers. I was very happy to hear this class, so we may judge better which way to go. In the long run the HP Virtual Connect Flex Fabric approach is not that clumsy as it looks within a superficial comparison to CISCO UCS. Honoring the caveats, the CISCO approach has to be well designed and even more carefully maintained that the HP one. Details will follow.

Configure VRRP on hp networking E5400-Family

Posted on by
Reply

Configuring redundant gateway services for hp ProCurve ProVision based switches is not a very big miracle. Basically its about switching it on. VRRP, saying virtual router redundancy protocol is similar to HSRP from  CISCO or CARP from Open BSD. VRRP itself as of today is standardized in RFC5798 by the IETF and follows hp’s habit of using industry standard protocols.

Why is it not redundant routing ? Well since the VRRP feature is enabled on an per VLAN basis and even more, it only defines a redundant IP interface within the according VLAN. The actual routing is covered independently from this. Assuming that the routing is configured properly the failing over IP interface ensures that the routing can happen. Itself provides only a redundant IP interface, which could be used as routing gateway, so we name the function redundant gateway services.

Different to other approaches by other vendors, VRRP only provides pairwise redundancy where the virtual IP interface is the same than the according VLAN IP on the owning routing switch. This address is failed over to the backup switch, who has a second IP interface configured in the VLAN. This is necessary to check the proper operation of the primary interface.

The partnering happens based on a so called virtual router ID, VRID which is defined within the VRRP configuration. This enables administrators to configure even different redundant IP gateways within one VLAN, if static routing requirements have this need.

So configuration on the Master works as follows. First configure the proper VLAN IP address. Naming VLANs is a clever approach and helps in the long run.

vlan 10
name "production east"
ip address 192.168.0.1 255.255.255.0

Then the VRRP feature is globally enabled:

router vrrp

Then the actual redundant IP interface is configured within the according VLAN context:

vlan 10
vrrp vrid 1
owner
virtual-ip-address 192.168.0.1 255.255.255.0
enable

Be aware that the VRID context is independently activated on a per VLAN basis and enabled within each VRID definition.

On the backup routing switch the VLAN IP configuration as wel as the VRRP activation look pretty much the same.


vlan 10
name "production east"
ip address 192.168.0.2 255.255.255.0

router vrrp

Within the VLAN based VRRP configuration here the backup role is defined:

vlan 10
vrrp vrid 1
backup
virtual-ip-address 192.168.0.1 255.255.255.0
enable

Voila, redundant gateway interfaces should be available.

Especially the strict creation of pairs is different to other implementations. Often sets of interfaces may be created. As well the assignment of the virtual IP as the identical IP that the VLAN IP on the owning router is not necessarily something other vendors do the same way. Very often the virtual IP is one and the local VLAN IPs are two different dedicated ones.

Blade New World

Posted on by
Reply

Most computer vendors flood the marketplace with more or less sophisticated blade solutions. Remark: Blade solutions well differentiated from blade servers.

So here some musings on blades and why I tend to differentiate between basic blade servers and the more sophisticated approach.

Basic blade concepts primarily convince, considering all the same aspects:

  • Optimized footprint aka rackspace
  • Reduced cabeling
  • Energy efficiency
  • Virtualized installation media
  • Cooling efficiency
  • Central management and maintenance
  • Easy hardware deployment and service

To whatever extend the different breeds address the different issues, these primarily are seem to be the supperficial quality criteria and argued in many decission processes, missing the core scope of the discussion. These topics are covered in the mainstream publications and are measured as the grade of quality of the “solution”.

From my point of view this is completely out of scope.

The real value of blades is interconnect virtualisation together with the so called “stateless server” approach. No surprise that many vendors try to keep the discussion on the less important facts, since according to my labs and evaluations only three major vendors even understood the issue. Depending on their legacy obligations they have more or less radical approaches to reach the goal.

The goal is to generate a server personality, its identity in terms of technical aspects, dynamically by application of a so called service- or server- profile. This profile contains the different aspects of the individual server identity such as e.g.:

  • BIOS version
  • Other firmware version, e.g. HBA or NIC
  • WWNs for port and node
  • MAC addresses
  • Server UUID
  • Interface assignements
  • Priorities for QoS and power settings

Accroding to these profiles and derived from pools of IDs, MACs, WWNs the servers identity is generated dynamically and assigned during a so called provisioning phase. Then the blade server is available for installation.

This approach allows to “on the fly” generate a server personality, that serves dedicated needs as for instance an VMWare vSphere server or an database server. Furthermore if more servers from the same type are needed the profiles my be cloned or derived from ma template so that new rollouts are quick and easy. In case of failiure or desaster recovery the profiles may even roam to other, not yet personalized servers and asuming a boot from SAN or boot from iSCSI scenario failed servers are back in minutes, transarent to even hardware based licensing issues.

Derived from specially the need for flexible interconnect assignment the classical approach of dedicated Ethernet or Fibrechannel switch modules in a blade infrastructure is of no further use any more. The classical approach needs dedicated interconnects at dedicated blade positions which is exactly the limitation a service profile wants to overcome.

With converged infrastructure, the support of FCoE and data center bridging as well as the according so called converged network adapters, thi limitation has been overcome on the interconnect side. Here the interconnect is configured in an appropriate way to cover the server side settings and assigns dynamically different NIC and HBA configurations to the single blade. Even more the connections may apply QoS or bandwith reservation settings and implment high connection availability in a simplyfied manner.

Based on that far advanced and very modern hardware operation concepts are possible. Only blade concepts, that support the full range of this essentially decoupling from hardware and service role, deserve the name “solution”. Anything else is “me too”.

Some posts on howtos from my previous evaluation and installation projects will follow. Some readers may remember my old blogg 😉