In my job as Cisco’s Field & Sales CTO for Borderless Networks in the Cisco EMEAR Theatre, I have the privilege of working directly with many Cisco customers and partners. The majority of these folks are what you’d call “Technical Decision Makers” and CTOs. They’re the IT leaders who do the planning, the strategy, and work on the evolution of their infrastructure.
And frankly, in my 20 years in the IT industry, I have never witnessed such a perfect IT storm!
We are in the midst of a time in IT, where, for most organizations, the current megatrends are having a profound impact on the relevance of their IT. This is felt in both the infrastructure technologies as well as the solutions required to support those megatrends.
The Megatrends we are seeing are BYOD, fuelled by an unprecedented proliferation of mobile devices, the advent of the Next-Generation Workspace, Video, Software Defined Networking, the journey to the Clouds (with both the multi-tenant cloud infrastructure as well as the networks required to access the various types of clouds), and the Internet of Things (as a subset of IoE).
What is unique (and forms the IT storm I was speaking about above) is that these Megatrends are all happening at the same time!
Yes, they pose some risks and challenges, but above all, they represent a fantastic opportunity to position IT as truly business relevant. All the megatrends above are about making the business more agile, aren’t they?
Let me give you a few examples of what’s hot on the agenda of the customers I am working with, and the questions they’re asking:
- With several key applications moving to the clouds, how do our customers ensure security and application performance? What if they deploy for instance Hosted Desktop solutions such as Cisco VXI, how do they ensure pervasive security?
- How do they effectively address the mobility of their VMs between clouds? Or within a Cloud?
- What do they need to take into account when it comes to Hybrid Clouds?
- BYOD users are using wireless/4G devices to access video-enabled cloud- based applications, which are continuously offering higher video quality. This leads to bandwidth and service delivery constraints. How do they sustain the growth both of devices and video traffic with new capabilities such as new video codec, SDN for service delivery selection and so on?
- How do they cope with the proliferation of devices on their network? They are currently looking at their unified access strategy as well as the next wave of WiFi technologies and services (802.11ac, Location Services, …)
- What is the Cisco ONE strategy? And how can network programming capabilities such as onePK help our customers provide services to their business in new and better ways?
- How do they cope with the increasing complexity of their networks?
- Knowing IPv6 needs to de deployed now to cope with the wave of IT megatrends, where do they start?
Let me ask you:
- How are you currently dealing with those megatrends?
- What have been the implications to your architecture and your business?
- What are you currently prioritizing? Which technologies are you currently considering deploying in your network?
Many thanks, and stay tuned!
How are we dealing with it ? Probably we’ll end up with the usual solutions: overprovisioning and other workarounds. Like we are currently looking at using overlay networks for multi tenancy.
I wouldn’t call it a perfect IT-storm, I would probably add SH in front of that. As all these things people want don’t fit the networks we have. And with the current economy there might not be money to fix it either.
If you want you can read me going of on a tangent below. It’s just my personal wishlist of small changes which can have a large impact if they did get widespread deployment and I didn’t even mention the bigger things like IPv6 or DNSSEC.
I’m hoping for some real solutions to come to the market like:
– lower prices for 10 gbe;
– cheap silicon photonics;
– CoDel or CoDel-like queue management to fix bufferbloat
– SPDY or HTTP/2, SNI and the tweaks of TLS and TCP so we can leave to get good latency again: these include tweaks like Fast Open, TLS False Start, Proportional rate reduction for TCP;
– someone to come up with the right answer for overlay networks, instead of all these ideas people have now which only solve part of the problems and probably create new problems;
– MultiPath-TCP for offloading mobile traffic over local links like WiFi and for load balancing in the datacenter based on things like congestion control not some silly hashing functions;
– someone to come up with some real protocol improvements for WiFi because current protocols break down really fast when you get more users. The loss and the latency are really bad on most networks, there is a lot of improvement possible;
– maybe some TCP with Network Coding to be able to better distinguish between packet loss and congestion in TCP
But I doubt any of them will come to the market on time or maybe even in the coming years. Seeing as SSLv2 is still deployed on more than 25% of the top 200000 ‘secure websites’ I doubt anything will happen.
Personally I’ve suggested maybe we could do overlay networks over MultiPath-TCP to fix multipath and VLAN-exhaustion in the datacenter in one go, but because it is not connectionless it’s probably a bad idea… but I’d love to hear an honest opinion from someone knowledgeable on that.
Hello Lennie,
Many thanks for taking the time to comment, much appreciated.
Actually, your points bring more food for thoughts, and are touching on a couple of aspects I had not discussed above explicitly enough. Thank you for that !
1. We have to start to move from “building networks that here to last” to “networks built for change”. Because some past solutions (you highlighted a couple of them) won’t just work in the future in my humble opinion. A good example of that is the recent Cisco Unified Access that I am sure you have seen. We designed and developed very innovative solutions (asic, hardware, software) which will allow our customers to cope more intelligently with the wired/wireless convergence and the proliferation of devices. In that case, old solutions (more switches and access points without any intelligent policy engine) would not work moving forward. See : http://www.cisco.com/en/US/netsol/ns1187/index.html
2. The IT storm discussed above calls even more for architectural approaches. That is how we can build “networks for change”.In future posts, I intend to discuss specific examples of that.
3. “As all these things people want don’t fit the networks we have”. BYOD comes to mind, and there are other examples. Yes, networks need to evolve and wil haev to evolve. Thank you for mentioning IPv6, fully agree. Ultimately, I see it as an opportunity for IT departments in charge of corporate infrastructure to bring even more value to their business because as you said “people want it”.
By the way, with regards to one of the things you listed, you may be interested in the introduction of the Next-Generation 100 Gigabit CMOS-based Transceivers. See : http://newsroom.cisco.com/press-release-content?type=webcontent&articleId=1157237
Again, many thanks for your comments Lennie. Eric
I guess my comments can mostly be summarized as has become almost impossible to deploy new protocols on the public Internet.
Which is sad. Especially as I work at a provider.
In your own datacenter you have a lot more freedom of course.
I don’t believe we’ll be able to keep addressing the trends by tweaking
current networks and increasing capacity. Whatever the network size,
whatever the underlying protocols there will be users with increasing
needs, and network admins struggling to meet those, while reducing
complexity and saving costs.
We’ll keep increasing links capacity but not everywhere at the same speed.
We’ll move to 40GE and 100GE in the DC while typical branch site will
probably only have few Mbps while all applications (and probably the
entire OS) will move most of the time in the cloud. Having a big network
has never solved netadmins problems. I remember some time ago when I was
operation manager of a large backbone my team had to troubleshoot
performance problems while my network was largely overprovisioned. We knew
problem was elsewhere but we had not only to justify the network was
meeting SLA’s but also had to fingerpoint what was broken. Networks now
have to work closer to the application. And not just to provide typical
network metrics. I believe progresses Cisco is making on AVC (DPI,
performance metrics, application and routing optimization…) are addressing
exactly that.
Same for security, adding bigger firewalls won’t scale. Companies can
hardly manage their firewalls and security appliances today and I don’t
believe that current overlay solutions can survive when network boundaries
are becoming blurrier every day. New approaches should be used and
infrastructure must also do more than just transporting data while
expecting devices (firewalls, …) to do the job.
For example, being able to transport the connection context across the
network through the use of Security Group Tags helps simplifying security
while reducing inherent complexity. I even believe we need to do more in
synchronizing security elements. Reputation built for email addresses and
web pages could certainly be better used by network devices.
Last, I believe none of the solutions I could think of will now be adopted
if not simple as network complexity is increasing. SDN and cloud solutions
will greatly help here adding orchestration capabilities. And not just
inside the datacenter.