Main Page

The Linux SCSI Target Wiki

Revision as of 19:58, 15 December 2010 by Admin (Talk | contribs)
Jump to: navigation, search
Welcome to,
the standard multiprotocol Storage Target in Linux.
100 articles


LIO ( is rapidly becoming the main open-source multiprotocol Target for Linux. It provides fabric protocols for iSCSI and FCoE, with current integration of Fibre Channel HBAs, and planned support for InfiniBand (both SRP and iSER). It is replacing the Linux kernel engine of STGT and will be going upstream into Linux 2.6.38 (winter 2010 release) as the standard Linux multiprotocol target.

LIO has become the Linux iSCSI target of choice for many volume storage appliance vendors, including Netgear, QNAP, Data Robotics, Synology, etc., allowing them to achieve VMware ESX 4 and VMware vSphere certifications.

LIO is developed by RisingTide Systems, an IP SAN systems and software company located in the Silicon Valley. RisingTide has complemented LIO with RTSadmin, a storage management UI and API. In December 2010, RisingTide will be releasing the RTS Director, a distributed, highly available storage management tool with a unified namespace for the entire storage infrastructures.


  • High-performance, non-blocking, multithreaded architecture with SIMD support.
  • Measuered 38.4 Gb/s / >350k IOPS over 4x 10 GbE NICs on 20TB flash backstore.
  • CPU architectures: x86, ia64, Alpha, Cell, PPC, ARM, MIPS, etc.
  • Distributions (with out-of-tree builds): CentOS, Debian, Ubuntu
  • Platforms: PC architecture, Sony PS2 and PS3, and Technologic TS-7800
  • No single point of failure within either logical or physical connections.

Advanced SCSI feature set

  • Persistent Reservations (PRs)
  • Asymmetric Logical Unit Assignment (ALUA)
  • Protocol and intra-nexus multiplexing, load-balancing and failover (MC/S)
  • Full Error Recovery (ERL=0,1,2)
  • Active/active task migration and session continuation (ERL=2)
  • Thin LUN provisioning (UNMAP and WRITE_SAMExx)

Multiprotocol target

Storage media independence

Compatibility and certifications

LIO is compatible to Initiators on the following operating systems:
  • Microsoft: Windows Server 2003/2008/R2, Windows XP/Vista/7
  • Apple Mac OS X Leopard (via 3rd party initiator)
  • Linux: RHEL 4.x/5.x, SLES 10.3/11, CentOS, Debian, Ubuntu
  • Unix: Solaris 10, OpenSolaris, HP-UX
  • VMs: VMware ESX and VMware vSphere, Red Hat KVM, Microsoft Virtual PC, Oracle xVM/VirtualBox, Xen

LIO passed VMware ESX 4.0 and VMware vSphere 4.0 certifications with Netgear, QNAP, Data Robotics and Synology storage appliances. It also passes the full Microsoft Windows Server 2008 and Windows Server 2008 R2 Failover Cluster compatibility test suite.

Standards compliance


RisingTide Systems also provides Core-iSCSI, a high-end iSCSI Initiator, in order to address a number of known issues with the Open-iSCSI standard Linux Initiator.

Core-iSCSI is available on Linux and Windows, and it has been ported to a wide range of platforms and devices, including:

RisingTide ported OCFS2 onto the Nokia Internet Tablets on top of the Core-iSCSI Initiator.

High availability and clustering

LIO is from gound up designed to provide excellent integration with high-availability and cluster scale-out.
  • Deeply embedded high availability (Network RAID1)
  • Scale-out clusters via concatenating LUNs
  • Scale-out clusters via striping (1H2011)


LIO is complemented with three levels of comprehensive management tools and frameworks:

RTS Director

RisingTide Systems also provides RTS Director, a distributed, highly-available cluster management tool. It comprises a shell, active library and API. The active library and API provide an open, extensible platform with a unified namespace that manages complex functionality, such as high-availability and cluster striping. The shell offers location-transparent access to all objects in the SAN cluster. New functionality and devices can be added via plugin-modules.

The RTS Director provides true zero configuration. It is based on a fully symmetric, distributed architecture - there is no single point of failure, no cluster controller, no resource manager, no central database, etc. Nodes running the RTS Director automatically discover and join the cluster when coming up (including the first time).

The RTS Director will be available in December 2010.

See also

Personal tools
Google AdSense