Load balancing is performed at the DR group level and not the individual LUN. ** Another problem to consider is that, for example, number of LUNs might be dictated by other solutions like Continuous Access EVA: "HP Continuous Access EVA requires that all LUNs within a DR group be owned by the same controller. Of course, everyone should test solutions before making a final decision. ** * To make things more complex, Oracle actually prefers large LUNs. That is another reason to avoid having many small LUNs. Watch out for scan/startup/shutdown times when considering large number of LUNs. Commonly, set LUN sizes from 64 to several hundred GB *. Nevertheless, it is worth testing fewer/larger LUNs spread across as many physical disks as the configuration allows, based on the expected I/O throughput and service times. Although Oracle’s recommendation is to minimize the number of LUNs per diskgroup by having fewer larger LUNs, to minimize the LUN management overhead, experience shows that I/O performance is more dependent on the number of underlying disks the LUNs are spread across. Set up LUN sizes in powers of 2, so 32 GB, 64 GB, 128 GB, and so on. If SAN-based boot is implemented, avoid sharing HBAs for boot and application LUNs. To avoid this problem, HP recommends moving to Fibre Channel fabric (switched) topologies. Using Fibre Channel arbitrated loop topologies can result in long latencies during initial device opens when using more than 512 LUNs per HBA. The design of an HP-UX server must be based on: Business Continuity Plans … and meet the requirements as set in the embedded document: What is the cost of downtime per hour? December 2, 2018 Glorified personal experience to prove that I “know best” (rather the opposite).
#SUEDED AN DMESH COMMON PROJECTS MANUAL#
My way of Operations Acceptance Testing: December 2, 2018Ĥ This Document is Not: A replacement for HP’s official statements.Ī written manual to learn HP-UX and its design principles in detail. It is based on my 25-years practical experiences in Unix/Linux. December 2, 2018Īs a classical electronics engineer, I believe in proper planning process – measure three times before cutting. * * There are so many other issues to discuss, but my aim is to list most critical ones that affect all HP-UX server in general. This document highlights some of the design elements for a server as a single entity (without virtualization or clustering components).
Once good design principles are adopted, the physical work on building servers should be easy and fast. However, most parts of this document are applicable to PA-RISC servers too. The focus is on Itanium-based servers, as PA-RISC has become obsolete. Everyone has different opinion – why not help system administrators and architects stop implementing bad practices.
Even after 25 years of Unix experience I cannot claim I know everything! Dusan Baljevic Sydney, Australiaįrequent “abuse” of good design principles.
My humble attempt to summarise well-tested HP-UX 11i v3 most critical design principles at the present time. Presentation on theme: "HP-UX 11i v3 – One Personal Approach to Basic Build Design"- Presentation transcript:ġ HP-UX 11i v3 – One Personal Approach to Basic Build Design
0 kommentar(er)
