Thank you!

In my probably second to last post in this blog I would like to take the chance to say thank you.

Thank you to Kevin Shine and his team at RSM Tenon, who organized our time here in Belfast and cared about us. You did a really good job and made our stay here very comfortable and enjoyable.

Thank you to Laurence Roberts from City Resorts, who provided our accommodation and always solved our problems concerning the house as fast as possible.

Thank you to the IT department at Translink for letting me work on many interesting projects and gain valuable on-the-job experience working in a live environment supporting many users across multiple locations. You all made me feel very welcome and I really enjoyed working with you.

Thank you to afib and OSZ IMT for making this internship possible at all.

I hope I did not forget anyone; if I did: thank you, too!

Farewell

Yesterday it was the time to say goodbye to my colleagues at Translink. To be honest I was a bit sad I have to leave now. I really enjoyed working on all these interesting projects with the nice people at Translink.

On my last day I finished some paperwork (documentations, trainings records etc.) and handed over the running projects to my colleagues.

I also had lunch with my supervisor and two colleagues at Gallopers, a sandwich and coffee bar, where I had a traditional Ulster Fry.

Finally I tidied up my desk – well, at least I tried to – so that the next placement student from OSZ IMT, who is starting on Tuesday, finds a clean and neat desk on his arrival … but I didn’t succeed. 😛

Enhancing Network Performance … Again

Last week I boosted network performance at Translink again.

The core switches used to be connected via single 1Gbit/s links, whereas all other ports are also 1Gbit/s. Unsurprisingly they had problems with several connections traversing the core. So I decided to increase the bandwidth (and by the way improve redundancy) by configuring an 8-port EtherChannel between each two core switches.

Of course the redundancy could also be improved by just patching more links between the switches, but in this case there would only be one usable link at a time, because the Spanning Tree Protocol (STP) would block the remaining links to prevent a loop in the network. By configuring the EtherChannel the switches consider the eight ports of the EtherChannel as one logical port and load-balance the traffic over all links.

Domain Controller Upgrade

We finally started upgrading the domain controllers (DCs) here at Translink, which are currently running Windows Server 2003, to Windows Server 2008 R2. Yay! As I told you in a previous post, this is necessary for the Exchange 2010 migration to start.

On Thursday after lunch Rory, a server technician working for Northgate, came to our office to support us with the migration.

First we set up a temporary (virtual) DC, moved the services and copied the FSMO roles to it.

Now we could start upgrading the DCs. As Windows Server 2008 R2 doesn’t allow an in-place upgrade from Windows Server 2003, we had to do a complete new installation. So we demoted the first DC to a normal member server and then removed it from the domain. Now we could install the new operating system, get a cup of tea, do some basic configurations, join the freshly installed server to the domain and run dcpromo to make it a DC again. Of course we repeated this procedure for all DCs, one at a time.

Finally we moved the FSMO roles back to the DCs they were initially running on, moved the services … and Bob’s your uncle! 😉

In the end we left the office at approximately 19:30.

Enhancing Network Performance

A project I’m working on that will never end is the enhancement of network performance, as there is always something that can be optimized.

So I did a lot of documentation work in the past week to reveal the most important things to optimize. After exposing the most crucial networking issues I started eliminating them: patching some more uplinks to improve redundancy, reconfiguring devices and so on. Unfortunately I cannot provide any details about the network here, as these facts are confidential.

But anyway, every configuration change I make and every uplink I patch is a small step towards a network which considers the best practices I have learned, e.g. the hierarchical network design:

Cisco Enterprise Campus 3.0 Architecture: The Layers of the Campus Hierarchy

Extending the Cisco Heaven

Another project I’m currently working on is the rollout of Cisco routers at Translink’s depots and garages, which are connected to the corporate network via ATM. At the moment these branch offices all use the standard all-in-one devices provided by the ISP. These devices unfortunately do not support advanced networking functions like VLANs, DHCP forwarding or routing protocols, which are needed for centralized network management. So that’s why they evaluate the Cisco router rollout.

After doing some internet research I configured my test device, a brand new Cisco 887VA, for PPP over ATM. After a failed test last week and a refinment of the configuration I successfully tested the setup at a train station near my office on Wednesday.

Cisco 887VA-SEC-K9
Cisco 887VA-SEC-K9

Active Directory Upgrade

This week I was introduced to another project at Translink: Plans are to upgrade the Active Directory infrastructure from Windows Server 2003 to Windows Server 2008 R2. The project is realized in cooperation with the IT service provider Northgate.

One principal reason for running this project is that another project depends on this Active Directory upgrade: the migration from Exchange 2003 to Exchange 2010.

A project assistant from Northgate came to the office to talk with my colleagues and me about the hardware (physical and virtual) and software requirements as well as the necessary network (re)structuring.

Getting Started with Microsoft Hyper-V

Translink is about to start a new virtualization project. They decided to use Microsoft Hyper-V as the hypervisor. The virtualization cluster, which is built on HP servers consists of 4 nodes which each have 2 6-core Intel Xeon processors and 96 GB of memory.

The first thing to do was to mount the servers in the rack which was done by my colleague John and me.

John cabling the power supply units

After approximately 90 minutes of work and 2 papercuts from the damn HP packaging (I have never got injuries from Dell packaging) the job was done:

the 4 virtualization nodes

New Tasks, New Toys (Follow-Up)

As you can read in an earlier post, I had to set up a testing environment for Cisco Smart Install, which is now up and running.

The Smart Install director (the device which manages iOS and configuration deployment to the Smart Install clients) acts as a DHCP server and configures new switches coming factory settings with an IP address, default gateway and a TFTP server address (DHCP option 150). Right now the director deploys images and configuration files only to devices with specific MAC addresses, as I only want to configure my test client at the moment. It is also possible to filter deployment based on device model and type or IP address (if already configured).

Cisco Smart Install console output on client
Cisco Smart Install console output on client

After being upgraded and/or reconfigured by Cisco Smart install the switch is ready to be used in the corporate network without the need to be configured manually. This feature is also called zero-touch deployment. The Smart Install director console output shows the successful image and config upgrade process:

Cisco Smart Install Director console output
Cisco Smart Install Director console output

Network Documentation

As most of you employed in IT may know, there are always some projects you do „by the way“ (e. g. Oliver, René and Mateusz should know what I mean). 😉

I also got (another) „by the way“ project: document the whole Translink corporate network from core to access layer and from OSI layer 1 to 3. This includes all networking devices as well as every cable that connects two devices.

I think I will use a top down approach and start with the network core and WAN links between the sites and then care about the inhouse network site after site. Probably I will also have to develop a hostname scheme as not all devices in the network have unique hostnames yet.

I’ll tell you what I make out of it in later posts.

Have a nice weekend!