The CSAIL network connects all locations in the research areas of the Stata Center with gigabit Ethernet at a density of one port per 25 square feet of office space, for a grand total of more than 4,000 gigabit networks ports. The network backbone is 10-gigabit Ethernet with a 30-gigabit aggregation connecting redundant backbone switches. CSAIL has aggregate outside connectivity of 1 Gbit/s.
The wireless network in the Stata Center consists of 72 Meraki MR14 access points operating at 300 Mbit/s with support for IEEE 802.11b, 802.11g, and 802.11n standards, providing full coverage of all research areas of the building.
Currently, CSAIL has about 2,500 systems active on the network.
Storage summary (as of September, 2013)
|| Total space
|| 64 TiB
|| 45 TiB
| NFS (persistent)
|| 399 TiB
|| 231 TiB
| NFS (scratch)
|| 76 TiB
|| 21 TiB
Most TIG provided network storage is backed up by default, though exceptions can be made by request. Nightly backups are kept for one week and then merged into weekly sets which in turn are kept for one month and merged into monthly sets. Monthly tapes are kept indefinitely though we only guarantee recovery for one year due to changing hardware and software environments (if you ask sweetly we can usually go back a bit further). Additionally AFS and the persistent NFS have periodic snapshots that are kept online.
Main page: NetworkTopics
We provide support for the data network throughout the entire Stata Center with the exception of the Student Street and certain areas below ground. This is very different from most of the rest of campus, which relies entirely on MITNet.
We provide DHCP throughout the network, except in areas where research groups have opted to run their own DHCP server. By default, every blue-colored data jack in offices and open lab areas is "live", meaning that it is connected to the network, enabled, and will provide an IP address, DNS server info, netmask, and other necessary information via DHCP.
There is no need to register or do anything other than just plug your computer in with an ethernet cable and make sure that your computer is configured to use DHCP. We can provide an ethernet cable if you don't have one; see "How do I Get Help?" below for more info. In the case of a wireless network card, you only need to associate with the "StataCenter" network. This means you can be up and running on the network in just seconds. (In order to print or use certain other services, you may need to register your wireless device.)
Main page: EmailTopics
One of our most widely used services is our IMAP mail service. At your option, you may choose to have email that is sent to your CSAIL email address (@csail.mit.edu) delivered to our IMAP server, which has many features, including the following:
- well-maintained spam filters
- automated Bayesian spam/ham learning
- virus filtering
- easy-to-use server side custom filters
- vacation auto-responders
- webmail access from any browser
Main page: WebServices
We provide website virtual hosting for individuals via people.csail.mit.edu, and research project hosting at hostnames that you can specify. Our virtual hosting servers include support for most popular web site tools, like PHP and MySQL. If you need a service that we don't currently offer, just ask; it may already be in the works, but if not, we'll do our best to accommodate your request.
Main page: AccountsAndSecurity
The CSAIL computing infrastructure uses Kerberos V5 at the core for authentication. Each CSAIL user has a CSAIL.MIT.EDU "Kerberos Principal", which is a strong authentication credential that is built upon cryptographic techniques. Think of it as your passport to all of the computing and information services CSAIL has to offer.
Main page: FileSystemsAndBackups
Kerberos works in conjunction with AFS, which is an authenticated, global network filesystem (as opposed to unauthenticated file systems like NFS) to provide a high degree of fault tolerant, globally available file service that has a much more robust access control model than standard UNIX permission modes. You can even create your own user groups, without sysadmin intervention, and assign those groups specialized privileges to your files. AFS also has a built-in recovery feature called "snapshots" that allows you to quickly restore a file that you've mistakenly deleted.
Additionally, we do provide NFS filesystems for groups that have needs that are not addressed by AFS, such as high-performance computing. We encourage you to use AFS for the majority of your storage needs, however.
Main page: SoftwareTopics
We have site license agreements for many commonly used software packages, including Microsoft Windows, Microsoft Office, Adobe Acrobat, Matlab, and many others. We can provide you with media and license keys, as well as installation support if your research group doesn't have their own system administrator. For more information on what TIG has available, go to CSAIL's Software Distribution Center
We also provide mirrors of a number of popular Free and Open Source software projects
Main page: AudioVideo
Main page: PrintingTopics
This page provides information on how to configure your computer to use the CSAIL printing infrastructure. Print spooling is handled by a software package called CUPS, which runs on the CSAIL print spooler. Printing service is available for just about any operating system, since CUPS is backward-compatible with LPR (an older UNIX printing system), and all printers are available via Microsoft Windows printer sharing.
Email TIG or stop by 32-268 if you experience any problems with our printing services; that is, print job spooling, client configuration issues, printer jams, toner, paper, or other physical issues with printers.
Operating Systems Support
Main page: OperatingSystemsSupport
What we don't go into there is that we can extend our deployment and configuration management systems to provide groups with somewhat customized version of our standard Linux distro for desktop and or server use (and possibly different customizations for different sets of servers)
OpenStack Private Cloud
Main Page: OpenStack
is a free and open software suite for providing Infrastructure as a Service (IaaS) both for private clouds like we're implementing and large public clouds like Rackspace. We are currently using "Nova" compute services, "Cinder" volume service, and "Glance" image management and "Neutron" networking service. We are currently using the "Havana" release (aka 2013.2 released October 2013) on Ubuntu 12.04LTS "Precise" with KVM as the virtualization layer.
We are running primarily on hardware donated by Quanta Computer in the summer of 2011 with a total of 768 physical cores using Intel Xeon L5640 2.27GHz chips (this presents as aprox. 6,000 virtual cores) and over 3 Terabytes of RAM. This is delivered in 64 dual-socket hex-core systems ( 12 physical cores presenting as 24vCPUs) 56 of which have 48G of RAM with the remaining 8 having 96G of RAM each.
In Spring 2014 we added 10 Dell r420 servers. Each of these is dual socket 8 core E5-2450L chips and 192G RAM. This expansion added about 1200 virtual cores at the current oversubscription ratio and nearly 2T of RAM (our current bottleneck)
-- JonProulx - 19 Mar 2014
Virtual Machines For Research Groups
Main Page: VirtualMachinesForResearchGroups
Note over the course of the 2013 academic year this service will be merged into the OpenStack system described above.
TIG provides hosting of virtual machines to allow CSAIL research groups to execute specific research-related tasks, such as specialized or high-load web hosting that's inappropriate for standard TIG-provided WebServices. Generally these virtual machines are used for persistent, public-facing tasks.
Main Page: ClusterComputing
If your group is planning on purchasing dedicated compute resources
please contact email@example.com to see if your needs would be better
served by buying into our cluster system.
TIG has built out a flexible, reconfigurable
compute clustering environment which is able to serve both as a
lab wide batch queued cluster based on
Condor and also host private or
priority access nodes for groups who purchase dedicated hardware. The short initial proposal(PDF format) was made in August 2006 after a period of testing various cluster options.
The design, illustrated above, allows the integration or separation of three main components: the queuing systems, the virtualization system, and existing physical servers. The intersections show the areas discussed in this document. The non-intersecting areas suggest the possibilities outside the scope of this document, most notably the existing private group compute resources in their current stand-alone configuration, the possibility of virtualizing infrastructural services, and the possibility of incorporationg workstation idle cycle harvesting into the queing system.
-- JonProulx - 08 Aug 2007
TIG develops and maintains various custom productivity tools, such a:
-- Main.achen0 - 14 Aug 2012
- Peeps: The HQ personnel database
- The RA Tool: Used by PIs, PI staff, Fiscal, and HR for setting up Graduate Student Appointments