Virtualisation of common workloads is the norm, and virtualised networks will soon be too. Back in 2009 I took my CCIE R&S and at the time, I had the full resources of being a Cisco employee supporting me. Now however, I find myself more of a lone-wolf and in order to keep up with the new trends of virtualised infrastructure and software-defined/developed-everything I’ve decided to invest in a small virtual lab box.
Now – being that I live in a flat with no dedicated study/office space and a partner who (quite rightly) enjoys having an aesthetically-pleasing home, I’ve come up with a box which I can hide in a cupboard and it’s relatively low-powered and almost silent..
- Supermicro X10SDV-TLN4F
- onboard – Intel Xeon D-1540/1541 SoC (System on Chip) – 8c + HT
- onboard – Dual 1GE, and Dual 10GE Base-T LAN
- onboard – Dedicated IPMI interface
- 64GB DDR4 RDIMM (16GB x 4)
- 256GB M.2 PCI-e 3.0 x4 SSD (Samsung SM951)
- Two 1TB Seagate Barracude (ST1000DM00) slow disks
- 16GB USB stick (to boot from)
- All wrapped up in a Cool Master Elite 120 Advanced
All the kit is on order and hopefully I’ll have some updates as to the build and performance as the weeks go by.
In an article titled “Places the CCIE can’t take me”, Ethan Banks recently wrote that network engineers need more and more to be aware of ‘the complete stack’; in my eyes this means the compute, the storage, the virtualisation, the applications and the management.
I’ve been lucky in that I was introduced to VMware in 2002 – you know, before ESX and vSphere, when you still had to compile Workstation from source. So when Cisco dropped the UCS bomb in 2009, setting up vSphere wasn’t alien to me – I was one of a few network engineers who could understand the interaction between all the components. It was a good time to be an engineer!
This wholistic knowledge I’ve carried forward today; I am a network engineer at heart and will always start there but I talk to other engineers and customers about all the other components too; what are you running on the network, how is it hosted, what hypervisor or bare-metal OS are you using, what type of storage is it and how is it accessed and a 100 other questions that lead me to some idea of what is trying to be achieved.
In the last few years I’ve also started to ask the questions around managing infrastructure; what do you monitor and how? How do you control and backup configuration? These questions have been spawned from exposure to financial customers, where availability, integrity and latency are high on the agenda. Infrastructure engineers have been scripting configuration tools for years, but now application developers are trying to do it as well and they get called DevOps.
In the future, I think there’ll still be a need for the specialist engineers we have today; network engineers, storage engineers; compute guys etc – but they’re all going to need to understand a more about the wider picture than they do now. The scariest thing for me, recently; talking to a DC network guy who doesn’t know the damnedest about vSwitches and in the same five minutes a Nutanix engineer who didn’t know if he needed a port-channel for his vSwitch uplinks or not.
Well – I had my first proper introduction to Meraki last month by doing their 1-day CMNA course and I have to say, I was very impressed.
Here’s a company that have taken edge networking (wireless and access switching) and security networking and made it easy. Taken the complexity of the CLI out, made the UI intuitive enough and made the whole “crap, how do I do this” experience a thing of the past. Sure, the kit and dashboard doesn’t have the bells, knobs and whistles as Cisco gear but sometimes there’s just no need for that.
Being able to attached a bit of the kit to the network and have it almost self-configure and become instantly visible in the dashboard is a far cry from having to find a console cable and manually configure management not only on the switch itself – IP address, syslog server, SNMP server/strings, local credentials, RADIUS or TACACS – but also on each of those monitoring systems as well. Think of the time saved here when it’s all done, automatically, as if by magic..
Now, don’t let me deceive you here, there are actually some pretty neat and fairly complex things you can do around MDM profiles, client-specific profiles (with client-specific firewall and QoS) and site-to-site or client-based VPNs, but they are all made much easier. Not to mention that EVERY device in the Meraki offering has Layer 7 capabilities (which is totally crazy!!) and makes good use of it.
Anyway – don’t take my word for it – try it out for yourself.
Oh, and before anyone asks just how expensive it is.. don’t forget, the license includes all the support you’ll need, hardware replacement, and you don’t have to license any additional or third party monitoring tools, so go factor that into your TCO before you dismiss it.
Well, doesn’t time fly when you’re in a new job! I’ve finally settled down into my new role and been badged at “Technical Architect” – I’m not yet totally convinced that I’m there yet but it’s something I aspire to be.
I’ve done a lot of on-site consulting and design work in recent months and with a break in the work stack I finally have time to spending ‘solutioneering’ and more importantly going on training and catching up on today’s network technologies. Not that I was being left behind, I’ve still been watching twitter and reading blogs, but I haven’t been able to see theory in practice.
Recent weeks have been a flurry of vendor activities, and I hope to put a few thoughts to paper shortly for each:
– CMNA 1-day Training – learning the fundamentals of Meraki
– Cisco UCS / ACI Integration – a pilot course, but prompted a lot of discussion between engineers
– VMware NSX for Internetworking Experts Fast Track – pretty much says it in the title.
In the mean-time, I need to work out a solution for getting notes from Evernote into WordPress!
Based on and in partnership with ZenOSS – Cisco are releasing a new product called UCS Performance Manager. There’s a tech talk on Cisco’s website which, if you can get past the waffling at the beginning and get onto the screen demo, looks pretty good. Sure, it’s a cobbled ZenOSS, but the idea is good – it brings together a complete visual of the utilisation of UCS, something I haven’t see anywhere else. It can include not only UCS infrastructure (Fabs, interface utilisation, blade usage etc) but also probe external switching infrastructure as well as the virtualisation layer (currently vSphere or Hyper-V).
I’ve been looking for a good training lab solution that doesn’t involved having a small office humming with old ISRs and Catalyst switches.. Having worked at Cisco, I was aware of the various internal options (IOU, Titanium) as well as the more widely available ones (GNS3). But now, Cisco have finally realised that not everyone can afford to build labs full of kit and are releasing a few products to support individuals and companies who want to test configurations and network designs. This isn’t new news (we’d heard rumours for over a year of a product called VIRL, Virtual Internet Routing Lab) – but I’m not sure everyone’s found all the pieces yet.
Cisco Modelling Labs – is intended to be a corporate solution to support designing and planning of routed networks and their configurations. It’s a fully supported product that needs some serious hardware to run on, but allows you to build a routed network in a simulated environment, configure all the components up and see how they behave. Currently they’re supporting IOSv (a virtualised version of IOS), IOS-XRv and the CSR 1000v – which pretty much covers your main routing OSes.
onePK – is a development kit designed around Cisco’s onePK. The ‘all-in-one’ VM is configured to provide three routers running IOSv, all interconnected and ready for playing with onePK Python and Java interfaces. You can however, reconfigure it to provide additional IOSv instances, as demonstrated here.
There is also a Beta programme for a /dev/inovate lab – however I can’t see what the cost implications of this are. It looks ideal for those intending to do some hard-core software/API development against Cisco’s gear.
I left Cisco back in 2011 to go contracting, and I promised myself “to do a few years” and see how I got on. The experience has been eye-opening, to say the least. I’ve had some up’s and downs, both professionally and personally during this time and I think it’s hardened me up a little for the better.
I’ve seen how some businesses are well integrated, have great processes and work hard to keep business continuity. I’ve also seen total calamities. In the process I’ve been exposed to other Vendors’ kit (and some of it’s pretty darn good!) as well as the Cisco-Customer relationship that, at times, can be fretful in places.
In all, I’ve enjoyed the last few years of being outside the Cisco bubble. I’ve made great use of my skills, learnt plenty new ones, and met some great engineers and designers. But now it’s time to work on the next phase of my career progression – I’ve always wanted to become a “Solutions Architect”. Yes, I’ll admit it’s a bit of a fluffy title, but it’s the one I’m after. I’m going to be joining a VAR (Value added reseller) next week and developing a role where I can work on both the high level and the low-level of building end-to-end network solutions.
I recently had a conversation with a new acquantance around instagram.. She didn’t like the idea of Instagram. With all the effort that photographers and artists put into Photoshop edits, she thought the idea that you could just apply a fancy affect to a random photo was in effect, cheating.
I can see her point – but I like to think of Instagram as the modern-day Poleroid. In fact, I’ve made a photo album with the first and last pages being a scattering of Instergrams and it worked to good effect.
So it turns out that the Juniper SRX logs every traffic flow through it, which is pretty handy if are just about to utter the words “I blame the firewall” – so rather the prosecute without evidence, hold off judgement for a for short cli commands..
This little tit-bit is aimed around SRX’s with Logical System’s configured – but it probably works just as well without them.
Firstly, drop down into the Linux shell at the command line:
jim@my-juniper02> start shell
All the traffic logs for a logical system are located in /var/logical-systems/<name of lsys>/log
% cd /var/logical-systems/my-lsys/log/
The traffic flows are stored in plain text and the logs are rotated and compressed over the period of a few days. That said, it’s easy to grep the entire collection for what you’re looking for, in this case we’ve got a little DNS problem:
% grep “/53” traffic | grep “DENY”
Aug 20 20:00:57 my-juniper02 RT_FLOW: %-RT_FLOW_SESSION_DENY_LS: Lsys my-lsys: session denied 192.168.63.6/50627->192.168.16.11/53 junos-dns-udp 17(0) rule-global-deny(global) zone-ad-clients zone-ad-servers UNKNOWN UNKNOWN N/A(N/A) reth1.63
Aug 20 20:01:02 my-juniper02 RT_FLOW: %-RT_FLOW_SESSION_DENY_LS: Lsys my-lsys: session denied 192.168.63.6/44644->192.168.16.10/53 junos-dns-udp 17(0) rule-global-deny(global) zone-ad-clients zone-ad-servers UNKNOWN UNKNOWN N/A(N/A) reth1.63
So there you have it. Before allowing someone utter that all-too-commonly used phrase again, try this simple bit of detective work.
Well, it’s July 2012 already, which means in January or before I need to re-certify my CCIE. I’m hoping they’ll have released the CCIE-DC written by then (I missed the chance to try the beta) and I can use that as the next step.. until then, I’ll write a few refreshing blog entries.
Sometimes I still can’t believe I have this thing.. I know it’s not held in as high-a-regard as it used to be, but it’s still pretty great to know the time and effort that went into it and the achievement of passing..