Serviceguard Extended Distance Cluster (EDC) with VxVM/CVM Mirroring on HP-UX, May 2008

SmartSync Recovery Accelerator

For certain applications/databases, VxVM offers a feature which is called SmartSync Recovery

Accelerator. If supported by the database vendor, the SmartSync feature uses an interface between

VxVM volumes and the database software to avoid unnecessary work during mirror

resynchronization. In an Oracle environment, VxVM/CVM and Oracle exchange information about

which volumes are used as online “redo-log” volumes. For those volumes, DRL should be set to

sequential. All other volumes should be configured without DRL, because Oracle uses the redo-log

information to recover the data volumes/files. Details about this feature differ by volume manager

release and have not been evaluated before writing this document. The VxVM Admin Guide provides

more information on this feature.

FastMirrorResync (FMR)

FMR maps are located in version 20 DCO volumes. Each data volume has one DCO volume attached

to it. There is one FMR map per detached plex and snapshot taken from a volume. If no plex is

detached or no copy of the volume snapped off, then write requests do not trigger I/Os to the FMR

map. This is the case during normal operation of an EDC. If a plex is detached or snapped off, the

CVM master node updates the FMR maps for all volumes.

In an EDC, FastMirrorResync helps speed up the recovery from inter-site storage connectivity failures

that cause plexes to be detached from volumes. Since all changes to a volume with a detached plex

are recorded in the FMR map, only the data that changed since the plex was detached needs to be

resynchronized when inter-site storage connectivity is restored.

If you expect inter-site storage connectivity to be unavailable for a long time—long enough for almost

all data in a volume to be changed (and in need of resynchronization after the link becomes

available again)—the value of FMR diminishes. If the FMR map lists most of the volume as needing

resynchronization, not much is gained by using FMR. Switching FMR temporarily off with the vxvol

command, eliminates the I/Os that update the FMR maps. This can help applications that are close to

being maxed out on physical write I/Os. Without FMR, a volume needs a complete resynchronization

when a plex becomes available after restoring the inter-site storage link.

Figure 7. Example vxassist command to create a 10 GB volume and vxprint to see the volume’s objects

$ vxassist -g dgEDC make vol1 10g \

layout=mirror nmirror=2 mirror=enclr enclr:DC1-XP12k1 enclr:DC2-EVA8k1 \

logtype=dco dcoversion=20 ndcomirror=2 drl=on fastresync=on

$ vxprint -vh

Disk group: dgEDC

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0

v vol1 fsgen ENABLED 10485760 - ACTIVE - -

pl vol1-01 vol1 ENABLED 10485760 - ACTIVE - -

sd DC1-XP12k1_2-01 vol1-01 ENABLED 10485760 0 - - -

pl vol1-02 vol1 ENABLED 10485760 - ACTIVE - -

sd DC2-EVA8k1_6-01 vol1-02 ENABLED 10485760 0 - - -

dc vol1_dco vol1 - - - - - -

v vol1_dcl gen ENABLED 856 - ACTIVE - -

pl vol1_dcl-01 vol1_dcl ENABLED 856 - ACTIVE - -

sd DC1-XP12k1_3-01 vol1_dcl-01 ENABLED 856 0 - - -

pl vol1_dcl-02 vol1_dcl ENABLED 856 - ACTIVE - -

sd DC2-EVA8k1_2-01 vol1_dcl-02 ENABLED 856 0 - - -