Revision as of 21:13, 27 March 2014

Support LVM on a volume attached to multiple nodes

Related blueprints

Goals

The goal of this blue print is to support LVM on a volume attached to multiple nodes.

Overview

Current LVM driver only supports iSCSI topology and a guest instance on nova compute is connected to a LV on a volume-group via iSCSI target.

In FC storage environment, when a volume on the storage is attached to multiple servers such as cinder node and compute nodes, these nodes can access a single volume-group on the volume simultaneously. So every LV on the volume-group is visible on each node.

Using this configuration, a guest instance will be able to transport data through a guest instance to LVs on cinder volume via FC directly instead of iSCSI target.

(a) Existing iSCSI LVM driver

(b) Suggesting LVM driver

Benefits of this driver

Better I/O performance and latency for guest instances
Less network traffic of data transfer between cinder node and compute nodes
Reduce work load of cinder node
Basic features of cinder such as snapshot and backup etc. are appreciable from existing cinder LVM driver.

Basic Design

Multiple attached volume

This driver uses LVM on a volume attached to multiple nodes. By using LVM on multiple attached volume, cinder node and compute nodes can access access a single volume-group on the volume simultaneously. So every LV on the volume-group is visible on each node.
A volume-group on multiple attached volume is necessary to prepare previously. The driver creates LV as a cinder volume from the volume-group.

Exclusive access control for metadata

LVM holds management region including metadata of volume-group configuration information. If multiple nodes update the metadata simultaneously, the metadata will be broken. Therefore, exclusive access control is necessary. Specifically, operation of updating metadata is permited only cinder node.

The operations of updating metadata are followings.
- Volume creation
  - When Cinder node create a new LV on a volume-group, metadata of LVM is renewed but the update does not notify other compute nodes. Only cinder node knows the update this point.
- Volume deletion
  - Delete a LV on a volume-group from Cinder node.
- Volume extend
  - Extend a LV on a volume-group from Cinder node.
- Snapshot creation
  - Create a snapthot of a LV on a volume-group from Cinder node.
- Snapshot deletion
  - Delete a snapthot of a LV on a volume-group from Cinder node.

The operations without updating metadata are followings. These operations are permitted every compute node.
- Volume attachment
  - When attaching a LV to guest instance on a compute nodes, compute node have to reload LVM metadata using "lvscan" or "lvs" because compute node does not know latest LVM metadata. After reloading metadata, compute node recognise latest status of volume-group and LVs.
  - And then, in order to attach new LV, compute nodes need to create a device file such as /dev/"volume-group name"/"LV name" using "lvchane -ay" command.
  - After activation of LV, nova compute can attach the LV into guest VM.

Volume detachment
- After detaching a volume from guest VM, compute node deactivate the LV using "lvchange -an". As a result, unnecessary device file is removed from the compute node.

Permitted operation matrix
Operations	Volume create	Volume delete	Volume extend	Snapshot create	Snapshot delete	Volume attach	Volume detach
Cinder node	x	x	x	x	x	-	-
Compute node	-	-	-	-	-	x	x
Cinder node with compute	x	x	x	x	x	x	x

Prerequisites

Use QEMU/KVM as a hypervisor (via libvirt compute driver)
Shared block storages between Cinder node and compute nodes via iSCSI or Fibre Channel
A volume group on a volume attached to multiple nodes
Exclude target scsi device from lvmetad on compute nodes
- When compute node attaches created volume to a virtual machine, latest LVM metadata is necessary. However the lvmetad caches LVM metadata and this prevent to obtain latest LVM metadata.

Configuration

In order to enable LVM multiple attached driver, need to define theses values at /etc/cinder/cinder.conf

Example

[LVM_shared]
volume_group=cinder-volumes-shared
volume_driver=cinder.volume.drivers.lvm.LVMSharedDriver
volume_backend_name=LVM_shared

@@ Line 1: / Line 1: @@
-= New LVM-based driver for shared storage in Cinder =
+= Support LVM on a volume attached to multiple nodes =
 == Related blueprints ==
@@ Line 6: / Line 6: @@
 == Goals ==
-* Implement a new LVM-based driver which supports a shared or clustered LVM on a SAN volume instead of using a software iSCSI target.
+* The goal of this blue print is to support LVM on a volume attached to multiple nodes.
 == Overview ==
-At the standard LVM driver, Cinder creates volumes from volume group and Cinder provides these volumes using software iSCSI target. Compute node recognizes these volumes using software iSCSI initiator remotely. In this method, Compute node is not necessary to be connected to a backend storage which includes the volume group. Therefore cinder can provide volumes to Compute nodes easily. However, user can't access a volume via SAN because this LVM driver presupposes a volume access via Network using iSCSI targetd(or LIO).
+Current LVM driver only supports iSCSI topology and a guest instance on nova compute is connected to a LV on a volume-group via iSCSI target.
-In contrast, the new LVM-based shared storage driver supports a volume access via SAN using a shared or clustered LVM on a SAN volume instead of using a software iSCSI target. By using a shared or clustered LVM on a SAN volume, guest VMs on compute nodes can access their attached volumes via SAN instead of Network. This will improve I/O bandwidth and response of guest VMs.
+In FC storage environment, when a volume on the storage is attached to multiple servers such as cinder node and compute nodes, these nodes can access a single volume-group on the volume simultaneously. So every LV on the volume-group is visible on each node.
+Using this configuration, a guest instance will be able to transport data through a guest instance to LVs on cinder volume via FC directly instead of iSCSI target.
-=== (a) Standard iSCSI LVM driver ===
+=== (a) Existing iSCSI LVM driver ===
 [[File:StandardLVM.png|border|500px|Standard iSCSI LVM driver]]
-=== (b) LVM-based driver for shared storage ===
+=== (b) Suggesting LVM driver ===
 [[File:NewLVMbasedDriverForSharedStorage.png|border|500px|LVM-based driver for shared storage]]
 == Benefits of this driver ==
-# Improve I/O bandwidth and response of guest VM by supporting a volume access via SAN using a shared or clustered LVM on a SAN volume
+# Better I/O performance and latency for guest instances
-# Less network traffic of data transfer between Cinder Node and Compute nodes
+# Less network traffic of data transfer between cinder node and compute nodes
-# Basic features of Cinder such as snapshot and backup etc. are appreciable from standard cinder LVM driver.
+# Reduce work load of cinder node
+# Basic features of cinder such as snapshot and backup etc. are appreciable from existing cinder LVM driver.
 == Basic Design ==
-* This driver uses shared or clustered LVM on a SAN volume between Cinder node and compute nodes. By using shared or clustered LVM, multiple servers can refer same VG and LVs.
+=== Multiple attached volume ===
-* This driver creates LV from VG on the SAN volume for guest VM. A VG on a SAN volume is prepared previously.
+* This driver uses LVM on a volume attached to multiple nodes. By using LVM on multiple attached volume, cinder node and compute nodes can access access a single volume-group on the volume simultaneously. So every LV on the volume-group is visible on each node.
-* LVM holds management region including metadata of LVM configuration information. If multiple servers update the metadata at the same time, metadata will be broken. Therefore, it is necessary that operation of updating metadata can permit only Cinder node.
+* A volume-group on multiple attached volume is necessary to prepare previously. The driver creates LV as a cinder volume from the volume-group.
+=== Exclusive access control for metadata ===
+* LVM holds management region including metadata of volume-group configuration information. If multiple nodes update the metadata simultaneously, the metadata will be broken. Therefore, exclusive access control is necessary. Specifically, operation of updating metadata is permited only cinder node.
 * The operations of updating metadata are followings.
-** Volume create
+** Volume creation
-*** When Cinder node create a new LV on a VG, metadata of LVM is renewed but the update does not notify other compute nodes. Only cinder node knows the update this point.
+*** When Cinder node create a new LV on a volume-group, metadata of LVM is renewed but the update does not notify other compute nodes. Only cinder node knows the update this point.
-** Volume delete
+** Volume deletion
-*** Delete a LV on a VG from Cinder node.
+*** Delete a LV on a volume-group from Cinder node.
 ** Volume extend
-*** Extend a LV on a VG from Cinder node.
+*** Extend a LV on a volume-group from Cinder node.
-** Snapshot create
+** Snapshot creation
-*** Create a snapthot of a LV on a VG from Cinder node.
+*** Create a snapthot of a LV on a volume-group from Cinder node.
-** Snapshot delete
+** Snapshot deletion
-*** Delete a snapthot of a LV on a VG from Cinder node.
+*** Delete a snapthot of a LV on a volume-group from Cinder node.
 * The operations without updating metadata are followings. These operations are permitted every compute node.
-** Volume attach
+** Volume attachment
-*** When attaching a LV to guest VM on a compute nodes, compute node have to reload LVM metadata using "lvscan" or "lvs" because compute node does not know latest LVM metadata. After reloading metadata, compute node recognise latest status of VG and LVs.
+*** When attaching a LV to guest instance on a compute nodes, compute node have to reload LVM metadata using "lvscan" or "lvs" because compute node does not know latest LVM metadata. After reloading metadata, compute node recognise latest status of volume-group and LVs.
-*** And then, in order to attach new LV, compute nodes need to create a device file such as /dev/"VG name"/"LV name" using "lvchane -ay" command.
+*** And then, in order to attach new LV, compute nodes need to create a device file such as /dev/"volume-group name"/"LV name" using "lvchane -ay" command.
 *** After activation of LV, nova compute can attach the LV into guest VM.
-* Volume detach
+* Volume detachment
-** After detaching a volume from guest VM, compute node deactivate the LV using "lvchane -an". As a result, unnecessary device file is removed from the compute node.
+** After detaching a volume from guest VM, compute node deactivate the LV using "lvchange -an". As a result, unnecessary device file is removed from the compute node.
 {| class="wikitable"
@@ Line 67: / Line 75: @@
 * Use QEMU/KVM as a hypervisor (via libvirt compute driver)
 * Shared block storages between Cinder node and compute nodes via iSCSI or Fibre Channel
-* A volume group on the SAN volume
+* A volume group on a volume attached to multiple nodes
-* Disable lvmetad on compute nodes
+* Exclude target scsi device from lvmetad on compute nodes
 ** When compute node attaches created volume to a virtual machine, latest LVM metadata is necessary. However the lvmetad caches LVM metadata and this prevent to obtain latest LVM metadata.
 == Configuration ==
-In order to enable Shared LVM driver, need to define theses values at /etc/cinder/cinder.conf
+In order to enable LVM multiple attached driver, need to define theses values at /etc/cinder/cinder.conf
 Example

Difference between revisions of "Cinder/NewLVMbasedDriverForSharedStorageInCinder"