There were several outages to report on:
Aug 8 (Friday) - Routing problems were introduced by misconfiguration of a router at UH Hilo. Many UH sites including the MKOCN experienced lost connections between 12:30pm and 2:45pm.
From Aug 28th to September 9th, the UH networks experienced several DoS attacks with similar symptoms. An interface on a core router would stop working after a brief DoS attack and a reboot was needed to reactivate. The problems was solved on September 9th with software patches from Cisco. These outages include:
"Although I am fairly confident that the wireless network at HP will have minimal effect at the summit, I am a lot more worried that inadvertent radio transmission from laptop computers at the summit could become a problem. This could have an insidious manifestation, such as only happening when a certain person observes! The installation of the wifi system at HP will surely encourage people to bring wireless enabled laptops to MK, and therefore could indirectly contribute to unwanted radio interference."
The Working Group agrees with the oversight committee that there is a potential problem with built-in transmitters of observers' laptops. However, the problem will occur with or without wireless implementation at HP because in the very near future, every laptop will be equipped with wireless capability. In the end, the Group decided that a recommendation should be made to the Directors to instruct observers to disable their wireless equipment before going to the summit.
For HP, there was general agreement to wait for the Cisco 802.11g product, which according to Cisco, will soon be announced. If that doesn't happen, field upgradable APs could be deployed with 802.11b radios which could be upgraded to 802.11g when the latter become available.
For the dormitories, Pui Hin is also looking into using dual radios: 802.11a for building-to-building connections and 802.11g for users connections.
Our goal is to negotiate deploying of new technology such as Dense Wavelength Division Multiplexing (DWDM) at MK that will eliminate layers of SONET equipment, making it easier to troubleshoot network failure relating to Verizon services and to lower the cost of connection especially at higher bandwidth. Further, the current SONET infrastructure has a maximum capacity of 11 1/2 DS3s. New equipment must be installed to provide additional bandwidth.
Verizon is currently considering two alternatives, enhanced SONET and DWDM. Enhanced SONET is less expensive to implement at low bandwidth. Along with traditional SONET provisioning such as OC3, DS3, and DS1, enhanced SONET will support ethernet, fast ethernet (FE), gigabit and fractional gigabit ethernet (GE). However, it will not simplify equipment layers for most types of services. DWDM will support many more services such as point-to-point SONET up to OC192 and fiber channel with no Verizon equipment at the customer site. Further, DWDM requires only two fibers to implement a redundant ring while SONET will require four fibers.
To move forward, Verizon will need from us some commitment on future bandwidth needs and how soon we can convert to the new technology. Our costs will involve a non-recurrent charge (NRC). Monthly charge for a GE could be $6k or lower and a FE at $3k, which is the same as what we are paying now for a DS3. However, many of the observatories are separating one or two T1s from their DS3s for voice and video and may not be able to take advantage of the FE at the same cost. Also, it is hard for the observatories to commit to the new services without knowing what the NRC might be.
Pui Hin will continue to work with Verizon on these issues.
This is a follow up of our March meeting when we discussed the need to eliminate the outdoor fiber panel. We were hoping to save some money by having Verizon pull new fibers into the IfA building at the same time they relocate SMA's DS3 to the Research Park. At the meeting, we learned that since SMA's DS3 uses Verizon owned fiber, it will not make any price difference for us to install private fibers at the same time.
However, the Working Group recognizes the need to relocate the fiber distribution panel. Due to budgetary constraints, it would be easier if we were to plan our requirements and costs for implementation for the next fiscal year. So far, Pui Hin has heard only from Gemini.
David Lapsley said:
"Ideally, I would like to transfer the data as fast as possible. I believe 45 Mbps is the current bottleneck limit to Mauna Kea. What would be the maximum rate you would be comfortable for sustained transfers (say 4 hours at a time) over these two days?"
There were no objections to the request and Pui Hin will work with the observatories to determine what will be the best time and maximum bandwidth usage.
Nick Johnson (through email) has the following suggestion for improvement to the topology of the MKOCN.
Pui Hin explained that due to the shortage of fiber within UH-Hilo, it will not be possible to connect each Hilo site with UH-Hilo. However, once new fibers are pulled into the IfA building, it will be possible for each Hilo site to have connections to two different routers in the IfA building where dual routes to UH-Hilo already exist.