PSU VoIP

Aside from touring the expo floor, I attended a few more sessions Thursday and then again this morning, gravitating toward the virtualization sessions. As we're now implementing virtual servers in our machine rooms, a lot of the material was review, but not all. I learned that Microsoft has a free hypervisor offering called Hyper-V. It doesn't have feature parity yet with VMware but since they're offering it for free, I would expect VMware to respond by lowering their prices on their infrastructure components. This would be very helpful for us. Having been psyched up by these virtualization sessions, I'm really looking forward to our group getting the first few servers moved to VM and machines turned off.

On the show floor I spent a little extra time talking to network monitoring/management companies and a DNS/DHCP/IP address management company. The latter gave me a blue hat with a cat logo. Guess who? We have honed our DNS and DHCP processes over the years with scripts and mastery of vi but there are better approaches.

I appreciated the breadth of the conference and being able to sample training in various areas. It was an exercise for my brain to be continually switching gears (going from session to session or booth to booth).

Future of the Voice Endpoint

Panel: Representatives from Avaya, Microsoft, and Siemens

There was some discussion on extensions made to SIP by vendors in order to fill in some gaps in the protocol. All acknowledge that SIP lacks what is expected by the customer. To Office Communicator/Server, Microsoft adds strong authentication (Kerberos and TLS) and implements SRTP by default. Avaya (like Cisco) extends SIP to match features that its native protocol has. Sending individual digits immediately to the PBX as they are dialed is another example. (KPML is an extension that allows this.)

Will soft phones replace desk phones? These guys in the industry say "no;" Avaya says fewer than 10% will ever abandon the desk phone for a soft phone; it's most useful in conjunction with a primary desk phone. People just want to pick up the always-on device and dial the digits. This matches with my own personal experience using soft phones (home & trial at work). Again, no one says that soft phones aren't useful (especially me--I love the idea) but they don't, and may never, stand on their own.

Enterprise 2.0: Evaluating the current "2.0" technologies

Blogs, wikis, social networks, tagging, mashups, modern portals. A good review of what's out there and some commentary on the usefulness of each in the enterprise. Implement them with a purpose, not just because they're the current thing. Confluence was highlighted as a good business-oriented wiki for its overall usability, file sharing and access controls. (Kudos to those who selected it for use at Penn State!)

Part 1: IBM and Web 2.0 / collaborative technologies. Presenter was using Lotus Symphony to present his slides; claims it's free to download and at least looks to be on par with Powerpoint.

Current iteration of Notes is slick and is incorporating the technologies that we are trying to smash together--directory services, blogs, wikis, social networking (Face/Space, Twitter), IM, the whole works. Still looks a bit like a clunky IBM product.

Look for a beta of their new Web 2.0 product, Bluehouse. Demo wasn't clear whether this was the name of the suite or a component.

Part 2: Cisco and virtualization. Why virtualize? (We're talking about virtual presence, and virtual servers/VM.) Costs, large campuses, multiple locations, globalization.

Speaker is promoting virtualization as a way to multitask more efficiently (argh!). Need to upgrade human brain first.

Now getting into a real issue about virtualization: IT silos and individual resource pools make it tough to get buy-in. Q: Would PSU be able to bring computing resources together out of the individual departments/silos? PSU Cloud in the future?

In many virtual server setups the weakest link is the hypervisor. Admins leave big security holes here. Beware!

Virtualization shifts mindset from server-centric to services-centric. A good shift.

Today I attended an all-day training session with the aforementioned title. I already have some experience using open source tools: we use Smokeping, Cricket/RRDTool, Multicast Beacon, and others. I have some experience with Wireshark. The value in today's training was hearing an experienced network professional (Mike Pennacchi of Network Protocol Specialists) talk about how he uses these tools. Understanding the concepts and seeing an expert use the tools in certain ways is more helpful than just reading the documentation.

Aside from an extended session on Wireshark, which really helped me get a better grasp on its usefulness, the speaker presented a few other tools that were new to me. nTop can use Pcap or Netflow data to describe network usage. Nagios works well on its own but becomes a super power when combined with Centreon. I'm not trying to start a link farm here. These links will be useful when I get back to the office and find some time to start digging in.

Because this was a session on open source tools, there was a brief discussion about the "support" issue: who supports open source tools? One comment that I liked is that, even though there's nobody on the line to yell at, with most large open source projects, the answers you get from searching Google are better and faster than what you'd get from phone support, anyway. I hate to admit this, because it makes me the last stop of responsibility in troubleshooting, but it's pretty accurate.

I'll be attending Interop New York this week and hope to have some interesting notes to publish here. Interop offers workshops and conference talks on a wide variety of systems administration and integration topics, including VoIP. I've got plenty of diversity in my planner for the week; this is my opportunity to learn some new things outside of my usual focus. CIPTUG, which is all-VoIP (but not all-Cisco, never mind the "C" in the acronym), is coming up in November.

My last entry showed some background images that I politely took from the Department of Public Information and cropped and gently edited for use on the Cisco IP phones. These are not available (except as noted) on Penn State IP phones; they're just some ideas I had and loaded on our test system to demo to my colleagues.

I would like our phone users to have these background images. What's more, I would like phone users to have the option to buy, for an additional $60-100, a color-display version of their two-line or six-line Cisco phone. It doesn't do any magic tricks; it's just colorful. (That's not entirely true; there are some under-the-hood improvements, but definitely nothing magical.)

In light of significant server upgrades that I'm working on now, why would I waste my time playing with and promoting background images and color phones?

Let me start off with a word of self-defense: I'm not wasting time. I am just not focused enough to be able to work on server upgrades for months on end without engaging myself in some other more creative work. Server upgrades are boring, and we are trying to not be creative with these; instead, we want to follow the vendor's rulebook to avoid headaches later. I think this is a reasonable course of action. It's also the reason I haven't blogged much about the server upgrades. Want to know what we're doing? Details on www.cisco.com.

At the last two committee meetings with PSU VoIP technical collaborators ("VoIP Tech") I gave a show-and-tell of background images and the 7965 color IP phone. Response from the team was not positive. Reasons focused on internal issues of inventory management, firmware tracking, and philosophical disagreements concerning our vendor and the future of the IP phone....

No sour grapes yet; I want a less-technical opinion on whether these are good ideas or not. Marketing folks might see these in a different light. So might end users, who don't care about our internal issues. I think our end users would like us to offer them more products, more creative and interesting products. We are all humans and like shiny things. We watch color TV. 1080i-High-Definition-80-inch-diagonal color TV. We don't want the Model T. We go for the fast, good-looking car, if we can afford it.

Some--actually a majority--of the users don't care what phone sits on their desks. They will be satisfied or unsatisfied based on certain simple factors. Others, perhaps executives with expensive office furniture or athletics coaches who decorate their offices with fifty years worth of trophies, will want a colorful lion shrine to appear on their phone when they're not busy recruiting the next dean or football star. Or any phone user just wants a change of scenery. I'm talking about minor things that can please the customer--all while in the background, we tech nerds work to make the system capable, available, reliable, redundant, and all other service-oriented goals.

So the answer to "why" is that I think the end users will like it. And that should be enough reason to write about such things here on a blog and promote them internally as service & feature offerings. I think more focus needs to be given, while fully maintaining the infrastructure, to end-user experience. We keep telling ourselves that we're not a utility company. Therefore, let's get past the utilitarian.

Not to be outdone by the foaming soap dispensers that proudly display the Penn State logo...

Here are some ideas I had for background images for the 7941/7961:

Standard (currently available)	PSU shield logo (currently available)	Nittany Lion 1	Nittany Lion 2
Old Main tower	Allen Street gates	At Beaver Stadium	Honors College wall

Color versions of the grayscale images.  (This is what you'd see on Cisco's higher-end color-display phones:  7945, 7965, 797x.  We don't offer these, but they sure look cool.)

Standard	Nittany Lion 1	Nittany Lion 2	Elm trees
Old Main tower	Allen Street gates	At Beaver Stadium	Honors College wall

Photography credit: Penn State Department of Public Information

I've got questions; help me with some answers.

Is telecom boring? Is the public excited about anything in telecom besides the iPhone?

Is voice-over-IP no longer progressive? (I think I know the answer to that one.) What would make VoIP phone service interesting again? What features would make you want to use your VoIP phone service instead of your cell phone when the VoIP "phone" (whatever form) and the cell phone are right next to each other?

What should phone service cost? If it cost nothing to dial overseas, would you make more international calls? Is the price already so low that toll charges don't really matter?

Who's on first?

What kind of convergence/integration do university people really want? Voicemail by e-mail? E-mail by voice? Softphones on the computer and web browsers on the phone? More video capabilities?

Would most people be happy if we just provided a bunch of ringtones and background pictures?

"It's not easy being green."

My blog, if you read it from the web page, is green. That's entirely coincidental. Green happens to be a color I like; I'm not trying to make any particular statements with the color of my blog (except that PSU blue was getting a bit stale).

The focus of "green IT" is mostly on lowering the energy use and heat production of servers, through more efficient servers or a reduction in the amount of actual hardware. In the non-VoIP part of my group, we're digging into a virtualization project that should reduce 50 pieces of server hardware down to about four if everything works out perfectly. Even a five-to-one reduction would be fine. I'm excited about this project for a number of reasons; one of them is that I won't have to look at as much hardware in the server room. Fewer server problems should be diagnosed as hardware problems. And hardware problems won't be so catastrophic--just use some VM magic to move a server from failing hardware to working hardware. This "greening" of IT first involves putting down a lot of green (money) before we see any environmental benefit, which will be secondary to all the system management benefits, the main reason we're doing it all in the first place.

In a number of years, equipment rooms and power and cooling requirements will have shrunk but we'll still be committing all kinds of environmental crimes through our still-growing IT staff: driving internal combustion engines at $10+/gallon of gasoline. Two losses of green there: the environment eats our exhaust and the gas stations eat our cash.

Several years ago, Penn State ITS bought a number of efficient departmental vehicles for the staff to use to travel around campus. These four-speed one-seaters are human-powered and the only emissions are from the driver. Sadly, the ITS bikes have been seen around campus turning to rust. Some have been reclaimed and are in active use; Telecom Building has one that gets regular use and I have heard rumor that 300 W College uses one now too. This program could use some fresh support. Biking around campus is pretty easy and you can take shortcuts; there's no getting stuck in traffic or waiting for the shuttle.

I sit here writing a blog at 6 pm because I biked to work this morning, as I have done most mornings since early May. However, the weather (stormy at the moment) sometimes dictates my schedule. A worthwhile tradeoff, in my opinion: I'm saving money, getting into better shape, enjoying the outdoors, and not polluting. Several of my coworkers also ride to work. This is just one shade of a green IT department.

VoIP and close relative VCoIP (video conferencing over IP) come into play by allowing us to meet from wherever we may be. Phone calls are cheap, conference calling is easy, and video is available on every Mac laptop (and before long, on every laptop). Set up XMeeting on your Mac and participate in video conferencing, wherever you are. It even works through a VPN (somewhat lower quality but usable). Meeting face-to-face and handing out paper copies of documents is familiar and makes us feel like we're doing business; the greener IT department can do the same thing using video conferencing with side content (whiteboards, Power Point, wikis, etc.). When the personal interaction is the main goal, Penn State's videoconferencing services meet the need; for a meeting that needs a lot of electronic content delivery, Adobe Connect does a great job. We have the tools right at our fingertips.

Someday we'll get the hang of telecommuting and people will be able to telecommute more easily. I know other organizations--non-edu, anyway--have already figured this out, but it's still an awkward thing in higher ed. I heard an analyst say this at a VoIP conference last year, and she's right. Telecommuting should not be a goal for a greener IT department, but we can all use the concepts and technologies that are part of the telecommuting method to be more environment-friendly and economical in our daily business.

I'm answering this comment-question on the main blog because it's a new topic; also, because I'd be glad to hear more about this topic.

Richard Rauscher said:

Bill, i have a question that's been burning inside for a while -- why do we implement separate networks for voice and data? Why not just use 802.1p/q and DiffServ?

I've been through three VoIP implementations, 2 with Cisco Phones & Call Manager and 1 with 3COM NBX's. I've always kept the phones and data on the same network with no insurmountable problems.

thanks!

The answer to this question is entirely non-technical. There may be technical reasons also, but I believe the non-technical reason fully covers it. (Note: this answer reflects my point of view from working in operations and my understanding of decisions made before my employment in the department even began. It is neither comprehensive nor authoritative.)

Cisco VoIP service offered by TNS at University Park is fully managed by TNS -- phone, jack, cabling, LAN equipment, everything. TNS also offers fully-managed data LANs but departments are not required to subscribe to this management for their data networks, and many do not. The simplicity of having one "voice network" infrastructure, as opposed to some of the infrastructure being stand-alone voice and some being integrated with a TNS-managed data LAN, is a significant factor in service management.

Data and voice packets traverse the same core network--there is not a separate physical core network for VoIP. The separation in the core is entirely logical (IP addressing/subnetting and access controls).

Richard (and others): What functionality could be gained, in Penn State's situation, by creating more of a converged data/voice network? Thanks for your question, and I look forward to hearing more.