diff options
| author | John Wilkins <john.wilkins@inktank.com> | 2013-10-22 18:12:46 -0700 |
|---|---|---|
| committer | John Wilkins <john.wilkins@inktank.com> | 2013-10-22 18:12:46 -0700 |
| commit | b8d54cdf23554e0d705dab81e449104a78a49f34 (patch) | |
| tree | 5ea3eb6b6b35784d85a36eaa19ec1e9ee9b2e94b | |
| parent | 53486afa018a859d1c04c1aeaf7a2a697a9332da (diff) | |
| download | ceph-wip-doc-install.tar.gz | |
doc: Fixed typo, clarified example.wip-doc-install
Signed-off-by: John Wilkins <john.wilkins@inktank.com>
| -rw-r--r-- | doc/start/hardware-recommendations.rst | 21 |
1 files changed, 13 insertions, 8 deletions
diff --git a/doc/start/hardware-recommendations.rst b/doc/start/hardware-recommendations.rst index 4af68ba8072..c589301a435 100644 --- a/doc/start/hardware-recommendations.rst +++ b/doc/start/hardware-recommendations.rst @@ -339,8 +339,8 @@ configurations for Ceph OSDs, and a lighter configuration for monitors. Calxeda Example --------------- -A recent (2013) Ceph cluster project is using ARM hardware with low -power consumption and high storage density for for Ceph OSDs. +A recent (2013) Ceph cluster project uses ARM hardware to obtain low +power consumption and high storage density. +----------------+----------------+----------------------------------------+ | Configuration | Criteria | Minimum Recommended | @@ -360,12 +360,17 @@ power consumption and high storage density for for Ceph OSDs. | | Mgmt. Network | 1x 1GB Ethernet NICs | +----------------+----------------+----------------------------------------+ -The project enables the deployment of 36 Ceph OSD Daemons, one for each -3TB drive. Each processor runs 3 Ceph OSD Daemons. Four processors per -card allows the 12 processors in with just four cards. This configuration -provides 108TB of storage (slightly less after full ratio settings) per -4U chassis. - +The chassis configuration enables the deployment of 36 Ceph OSD Daemons per +chassis, one for each 3TB drive. Each System-on-a-chip (SoC) processor runs 3 +Ceph OSD Daemons. Four SoC processors per card allows the 12 processors to run +36 Ceph OSD Daemons with capacity remaining for rebalancing, backfilling and +recovery. This configuration provides 108TB of storage (slightly less after full +ratio settings) per 4U chassis. Using a chassis exclusively for Ceph OSD Daemons +makes it easy to expand the cluster's storage capacity significantly with +relative ease. + +**Note:** the project uses Ceph for cold storage, so there are no SSDs +for journals. .. _Ceph Write Throughput 1: http://ceph.com/community/ceph-performance-part-1-disk-controller-write-throughput/ |