Features

From OpenSSI

OpenSSI is a single-system image (SSI) clustering solution that extends the GNU/Linux operating system to address all categories of computer clustering: High Availability (HA), load balancing, and high performance computing (HPC).

An introduction to OpenSSI is available at our architect's corner.

Contents 1 Features Comparison 2 Base OpenSSI Features 3 Built-in OpenSSI Enhancements 4 Extended OpenSSI Features 5 HPC Middleware Support 6 Server Support 7 OpenSSI Limitations

Features Comparison

Matrix of Built-in Features

	OpenSSI	openMosix	Kerrighed
Single Administrative Domain	Yes	No	No
Single Install	Yes	No	No
Cluster Membership Guarantees	Yes	No	No
High-Availability (HA) Clustering	Yes	No	No
HA Shared Root Filesystem	Yes	No	No
Single Process Management Space	Yes	No	?
Process Migration	Full	Partial	Partial
Process Load Balancing	Yes	Yes	Yes
Process Migration is HA	Yes	No	No
Migrate Processes Using Semaphores	Yes	Yes	No
Migrate Processes Using Shared Memory	Yes	No	Yes
Single Thread Migration	Only thread group	No	Yes
Process Checkpointing	3rd party	3rd party	Experimental
Single IPC Namespace	Yes	No	?
Distributed Shared Memory	Migrates	No	Yes
Single PTY Namespace	Yes	No	No
Single-Site File Naming	Yes	No	?
Coherent Cache / File Access	Yes	?	?
Migrate Active Filesystem	Experimental (2.0.0pre2)	No	No
HA Cluster Filesystem (CFS)	Yes	No	No
HA Single Cluster Name/Address	Yes	No	No
HA Network Load Balancer (LB)	Yes	No	No
LB Auto Detects TCP/UDP IP Services	Yes (UDP 1.9.1)	N/A	N/A
Transparent Socket Migration	Not yet	No	Yes
Automatic Service Failover	Yes	No	No
Diskless Nodes via Network Boot	Yes (PXE/EtherBoot)	No	No
CFS support Direct Attached Storage	Yesdrbd	N/A	N/A
Support Lustre as Shared Root	Experimental	No	No
InfiniBand Cluster Interconnect	Experimental	No	No
64-bit x86_64	Experimental	No	No
Virtualized with Xen as DomU	Debian-SSI	No	No
Virtualized with Linux KVM	Yes (1.9.3)	?	?
Virtual Partitions / Guests	Not yet	No	No
Support NSA SELinux	Not yet	No	No
Zero Config LiveCD	Yes	3rd party	No
Support Multiple Linux Distributions	Yes (limited)	Yes	Yes
SMP (Multi-Core) CPU Support	Yes	No	?
Big Kernel Lock Preemption	Yes (1.9.3)	?	?
Preemptive Kernel	Yes (1.9.6)	?	?
Kernel Hot Updates	Not yet (2.0)	?	?
Support Thousands of Nodes	64,000 (3.0)	?	?

Note drbd: CFS soft mounts do not require DRBD. HA-CFS hard mounts require OpenSSI modified drbd-0.7 or vanilla drbd-8.2 starting with OpenSSI 2.0.0pre3.

Base OpenSSI Features

A summary of our feature list.

• Single management and administrative domain.
• Single root filesystem enforced across the cluster.
• Single copy of binaries, libraries, and admin files (like password).
• Single PID per process and a clusterwide process management space.
• Transparent, clusterwide namespace for all IPC objects.
• Clusterwide device access and a single pty namespace.
• Consistent "single site" file naming across all nodes.
• Transparent and fully coherent file access across all nodes.
• Integrated with most cluster filesystem technologies; provides flexibility and choice.
• Kernel interfaces to allow integrating other open source technologies.
• A highly available cluster filesystem with transparent failover.
• A single name and address for the cluster and that name/address is highly available, with persistent connections.
• Full process migration including its system calls.
• Exec-time process load balancing.
• A variety of high availability features including
  • process monitoring and restart,
  • automatic service failover,
  • automatic filesystem failover,
  • cluster ip address and connection management failover,
  • and the ability to lose a home node without killing all the processes that started on it.
• Strong membership guarantees and APIs for membership using Cluster Infrastructure for Linux.
• APIs for rexec() and rfork() as well as migrate.
• Diskless nodes via network boot.
• Features offered in mainline Linux kernel and GNU/Linux operating system.

Built-in OpenSSI Enhancements

Features that are included in OpenSSI release package.

• Keepalive - built-in availability management.
• HA-LVS - built-in load balancer with automatic failover for highly available IP address(es); based on LVS. Includes support for High Capacity Highly Available and Load Balancing and Streamline High Availability and Load Balancing topologies presented in Ultra Monkey.
• HA-NFS - built-in NFS server with automatic failover; parallel NFS client mounts.
• OpenDLM - built-in distributed lock management in OpenSSI cluster file system (CFS).

Extended OpenSSI Features

Third-party open source technologies that have been integrated or modified to enhance your OpenSSI clustering experience.

• DRBD - redundant fault-tolerant logical block device integrated for highly available shared root in OpenSSI without shared storage. (OpenSSI-1.2 and OpenSSI-1.9)
• Infiniband - ultra high-speed network support, network boot over IB (IPoIB). (OpenSSI-1.2; SSI-1.9 IB under development)
• Lustre - ultra high performance filesystem integrated for highly available shared root in OpenSSI. (OpenSSI-1.2)
• OCFS - Oracle cluster filesystem.
• OpenAFS - parallel network filesystem. (testing on OpenSSI-1.2)
• XEN support - built-in support for virtual machine monitor for concurrent execution of multiple OpenSSI operating systems with resource isolation. (Debian-SSI)

HPC Middleware Support

Many of the open source High Performance Computing (HPC) middleware has been run on the SSI cluster, along with some purchasable capabilities.

• MPICH
• LAMPI
• HP MPI
• openPBS
• ScalablePBS
• SLURM
• ganglia
• supermon
• Totalview
• Lustre
• PVFS
• Maui

Server Support

Servers that have been "tested" in OpenSSI. Process migration was not tested.

• LTSP - Linux Terminal Server
• Apache 1.3, 2.0
• Jakarta Tomcat 4/5, BEA Weblogic Server 9
• MySQL Standard and Max with NDB Cluster, Sybase, PostgreSQL
• Sendmail, Postfix, Dovecot, SpamAssassin, ClamAV
• OpenVPN
• Samba, vsftpd
• BIND, ISC DHCP, ntpd, OpenLDAP
• Squid
• Bacula, rsync
• TightVNC, NX (NoMachine)
• Shoutcast, ICEcast
• Servers written in PHP 4, 5
• Servers written in Python
• Servers written in Java
• ...
• mostly everything that comes with your Fedora, Debian, etc. distribution
• anything that runs on Linux 2.4 or Linux 2.6

OpenSSI Limitations

• Cluster Membership: Maximum number of nodes per cluster. 125 nodes in OpenSSI-1.2. 15 nodes in OpenSSI-1.9 (requires recompile to support more). Thousand nodes with future proc hooks.
• Cluster Filesystem (CFS): Maximum filesize, files, directories inherit physical filesystem's.
• UID/GID: 16-bit in OpenSSI-1.2. 32-bit in OpenSSI-1.9.
• Major device types: 255 in OpenSSI-1.2. 4095 in OpenSSI-1.9.
• Minor device types: Maximum number of mountpoints per filesystem type. 8-bit in OpenSSI-1.2. 20-bit in OpenSSI-1.9.
• IPC: Maximum number of semaphores, shared memory segments, message queues. Adjustable up to 32,768 each (for the cluster).
• Sockets: Maximum number of sockets.
• Processes: Maximum number of processes. 32,000 in OpenSSI-1.2. 32,768 per node in OpenSSI-1.9.
• PTY: Maximum number of pty's.
• HA-LVS: Maximum number of connections, directors, CVIP's. Same as original LVS - millions of connections.
• DRBD-SSI: Maximum volume size per drbd device, maximum number of drbd devices. Same as original DRBD.