Home
http://blog.radler.jp/?p=226 written in japanese describes in detail QUADStor installation, setting up an iSCSI initiator on ESXi and accessing QUADStor VDisks as datastores and the performance benefits of VAAI cloning.
QUADstor Installation & Configuration http://t.co/DiXsA7Tq5t
— Mitsutaka Ohisa (@m_ohisa) May 3, 2014
This article in french describes the installation procedure and goes into the details of a mirroring setup such as fencing configuration with Dell servers.
Recently with the integration of offloaded data transfers (ODX) into the opensource code, we broke a main feature of our product. vStorage APis for Array Integration (VAAI) Full Clone/XCopy feature no longer worked. This breakage can be verified by using esxcli as shown below
Introduction
Most of our linux installation are over the network using pxeboot, while the FreeBSD installations are using a usb stick. The reason for this that our servers do not have a DVD/CD-ROM drive. Recently we had to install Window Server 2012 on one of these servers and had to face many issues. Firstly lets highlight the issues we have
Offloaded data transfers (ODX) support was added in release 3.0.53 of the storage virtualization software. As with VAAI, offloaded data tranfers enables high speed copy operations by offloading the entire copy operation to the array. This feature is beneficial for virtual machine (VM) clone operations, significantly reducing host CPU and network utilization. Users of ESXi will notice the similarities with VAAI third party / full copy primitive.
With release 3.0.52 of the storage virtualization software, a native OpenStack cinder volume driver is available. Prior to the availability of the driver, in order to use vdisks for cinder volume the procedure would be to create an LVM volume on the vdisk and then use the LVM volume when creating a cinder volume. This is simple in itself and in the absense of a cinder volume driver for a storage array this would be the preferred method. However features like volume snapshots will be at the LVM volume level rather than at the driver level.
http://www.v-front.de/2012/02/quadstor-delivers-storage.html gives an introduction to quadstor storage virtualization. This article although old is still relevant. Users of the storage virtualization software will notice a different GUI, and many features found the present software missing. At the time of writing this article the only features available were thin provisioning, data deduplication and VAAI support.
http://www.v-front.de/2013/11/testing-veeam-v7-tape-support-with.html describes testing Veeam Backup & Replication 7.0 product with tape devices. In version 7.0 of the popular data protection software for virtual environments, support for tape based backup including virtual tape libraries has been added.
http://www.virt-infrastructure.com/?p=110 shows the setup of QUADStor virtual tape library (VTL) with NetBackup 7.5.x. The article also describes the setup using QUADStor storage virtualization for data deduplication. While the article does not go into the details, the screenshots illustrates most of the necessary steps such as creating virtual disks, using virtual disks as storage for virtual tape cartridges, accessing the virtual devices from NetBackup and performing backups.
http://blog.tsugliani.fr/vsphere/vmware-lab-storage-software-appliance-q... describes QUADStor setup on Linux (Debian 7.1) with ZFS ZVOLs as physical storage. The article later describes the benefits of VAAI and illustrates the performance benefits of VAAI over a non-VAAI setup.
In this article we describe some of the common commands used for cinder volume configuration such as creating volumes, deleting volumes etc. Prior to using the command line examples you would need to source the required environment variables as described in http://docs.openstack.org/user-guide-admin/content/cli_openrc.html
This article assumes that you have mt installed on your system.
For RHEL/CentOS this can be installed using the following command
# yum install mt-st
For FreeBSD mt can be installed using the following command
# pkg_add -r mt
For other distributions refer their respective documentation
Tape devices usually have two device files a rewind and a no-rewind device. For example on Linux the first tape drive can be accessed using either /dev/st0 or /dev/nst0. Similarly on FreeBSD we have /dev/sr0 and /dev/nsr0. This article describes the difference between the two.
The Linear Tape File System utilities use a block size of 512 KB by default. When trying to format a tape using LTFS on a windows 2008 (possibly windows 2012) machine the formatting will fail and LTFS will complain that the maximum block size of the HBA adapter is less than 512 KB. To fix this problem do the following
Introduction
On unix systems the mtx utility is useful for media movement within a tape library. This article describes some of the most common commands for tape movement within a library. The mtx utility takes commands of the form
mtx -f <sg device> <command>
Often we have found ourselves having to manually remove a volume entry from the cinder database directly. This article describes the steps to work with cinder database.
In order to work with the command line, the required environment variables need to be setup as described in http://docs.openstack.org/user-guide-admin/content/cli_openrc.html
Listing the cinder volumes
First we get a list of volumes which we need to manually delete as seen below
Download lin_tape and lin_taped
Download the latest lin_tape source rpm and lin_taped binary rpm for your Linux distribution. In this article we are working with RHEL/CentOS and we have the following files downloaded
sh# ls -1 lin_tape* lin_tape-1.74.0-1.src.rpm lin_taped-1.74.0-rhel6.x86_64.rpm
Build lin_tape source
As superuser rebuild the lin_tape source rpm
Introduction to LTFS
Starting with LTO-5 drives a new technology called Linear Tape File System (LTFS) was introduced which allows for accessing tape media through a file-system interface. This is acheived by partitioning the tape into two, namely the data partition and the index partition. While the data partition contains the user data, the index partition can be thought of the partition containing metadata about the user data.