University of California, Berkeley
Network Advisory Committee (NAC)

July 19, 2004 Meeting Notes
Network Advisory Committee


NAC Attendees: Mike Pivonka,  Karen Eft, Paul Vojta, Tom Holub, Michael Sinatra, Jeanette Zerneke, Craig Lant, Michael Quan, Ryan Means, Peggy Flens, Tessa Michaels

CNS Staff: Cliff Frost, Peggy Huston, Mike Sawyer, Dat Le, Michael Green, Mike Hunter, Leslie Bowden

Announcements

Tom announced two new members to the group; Craig Lant from System and Network Security and Michael Quan from Public Health Information Systems.

The "Firewall Service" item on the agenda is postponed to the August meeting.

Update on AirBears

Q:  Will Airbears at the new Data Center cover the greenhouses on Oxford?
A:  We don't think so; there may be coverage but not by design.  If the greenhouses want coverage, they will need to pay for it.

Q:  Is there a policy on "pringle netting", aka directional antennas?
A:  We have not tried it yet but we may be soon.  We have a grant for some mesh networking. Cliff gave a brief tutorial on mesh networking explaining that traditional networking is access points with individual links to a network (i.e. islands of connectivity).  Meshing would set up access points where the coverage overlaps and one access point links back to the network.  Mesh networking type models have traditionally proven unreliable.  In theory, one access point could go down and not affect connectivity.  

Guest account signup for AirBears is working successfully.  Jeanett confirmed that they have used it.  The service is not in production yet until we get the "real certificate".  Hopefully within a month.

We now have an order entry for AirBears on the Shopping Cart for applicants that have received a grant from Stark.

Q:  Is there any agreement for guest accounts?
A.:  Karen checked out the agreement for Guest Accounts and has approved it.

Q:  Do we have a method for tracing a machine that has caused problems on the network?
A:  Yes.

Q:  Does the guest account agreement state one person per machine?
A:  I think so (Cliff).

Q:  Is there any work on improving the reliability at Haas?
A:  We are working on this; improving the topology network campus wide.  There are known issues in older installs.

Soda and Cory now have AirBears coverage throughout the buildings.

Restricted data minimums may affect the wireless access; there may be no sensitive data on laptops; email is an issue; and encryption should be used.

Q:  Is there an effort to encrypt all AirBears?
A:  We are looking at 802.1x.  We do not have control of the hardware so we can only provide a broad variety of encryption type.  It is up to the user to elect to use it.

Data Center Questions

The data center will host servers and services; video conferencing, high bandwidth, web casting, streaming videos.  IIS is already doing some of this.  ETS offers similar services.

Q:  What is the co-hosting structure going to be?
A:  Don't know yet.  Need to determine how to charge, e.g. network or space, high/low availability and redundancy needs, etc.   Customer would speak to one rep at CCS to coordinate agreement.

We are working on a tour of the data center for the NAC group in late August.  The library is currently planning to move their network to the Data Center.

Q:  Is there physical security, e.g. cages?
A:  Currently there is one big room with secured racks.  (Not sure if they are independently keyed.)  

Q:  If more pepole have access to the room, is the security lowered, i.e. someone could disconnect a machine from the network.

Q:  Is the intent that the servers will be housed in the Data Center but will be administered by the departments?
A.  Yes.

Network Funding

VCAC was being asked to endorse a model similar to voice at the last meeting.  Some people were upset because it would happen so quick.  VCAC wanted to consult and get feedback from campus.  The current plan is to implement a recharge model January 1, 2005.  Cliff has been charged with speaking with groups on campus.  Cliff posed the following questions to NAC:
1.  Who should we present to?
2.  What are the pros/cons?

We will get rid of the nodebank.  Departments will receive funding to increase base budget to the extent of the current databank.  Departments will receive a monthly billing and pay per node like the current system for phones.

Cons:

  • If a department exceeds the number of nodes they started with, they will pay full amount for additions.
  • Departments may use the funds to spend elsewhere  (not CNS), i.e. undermine uniform access to the network, uniform wiring, etc.  The result could be pockets of inferior infrastructure.
  • The allocation in the base budget may suffer from budget cuts in future years.
Pros:
  • Puts the decisions and funding in department hands
  • Cleaner relationship between CNS and customers.
  • Every department is treated the same; no exceptions (which make it difficult for CNS currently).
  • The funds being allocated to many departments may actually be allocated instead of current situation where CNS has not received anything.  
  • To accompany the distribution of administration, we will need to issue more standards, regulations, etc.  Network policing will increase to ensure that departments are doing what they are supposed to.
  • Departments will have a tough time prioritizing network nodes high enough.
  • How much damage will occur before the department figures it out?  Will they figure it out?
Nodes in the nodebank based on chartstring and type of funding department had available i.e. auxiliary or state funded.  Exceptions posted on CNS website (SBA's), e.g. Alumni House.

Q:  How much money will CNS loose?

Q:  At what point is it proper to do the transition?

CNS will need to educate campus.

The following names were mentioned in response to "who should we present to?"
ABOG
BASBOG
Micronet
Academic Senate
CAPRA
SISC

Next Meeting

August 9, 2004

ITAC home | NAC home  
Email about this website: nac-website@uclink.berkeley.edu
Updated: August 24, 2004