22 posts • joined 30 Mar 2011
SDN enables orchestration. The largest stumbling block in cloud orchestration is the network. It's all well and good automating your virtual switches, adding port groups and vlans within the hypervisor stack, but it's when you break out of that stack that things get interesting. Suddenly you are dealing with firewalls, routers, and switches which traditionally have not been designed with centralised automation in mind. It SHOULD be easy to programattically add another virtual firewall / router context / vlan etc on these components, but it's not, certainly not with traditional vendors. That's where SDN comes in, particularly in multi-tenant environments with different customers traversing the same tin or piece of copper. NFV is even more exciting. Why exit traffic from a host, through a switch, firewall or router, and then back to the same or nearby host to move traffic from vm to vm? It's crazy, and completely inefficient. With Intel baking IP packet management into their chips, which makes moving packets CPU efficient, a virtual router or firewall moving the traffic from VM to VM without traversing the external network just makes so much sense. And being a new-world software device, exposing an API for orchestration should be relatively simple. Unfortunately Cisco forgot that part with their 1000V series, but that's just typical of Cisco.
As a cloud service provider, no technology has ever had me salivating in desire for a mature, interoperable solution than SDN.
Aside from the orchestration capability, it also removes the most troublesome parts of running a cloud - network engineers.
Both the Austalian and the AFR have gone from respected national papers to little more than mouthpieces for Murdochs hard-right-wing idealogues. They are more likely talking to the current governments wishes than any reality.
More than just monitoring machines
Every coke machine down here in Oz is already network connected. They all have GSM (well they used to, I assume they have moved to 3G) antenna's that phone home to CCA with stock levels. They've been doing this for years and years.
I'd suggest maybe this is for something bigger like a global RFID deployment or maybe a home wifi connected postmix dispenser with automatic re-ordering?
I lay part of the blame at EMC's feet ;-) But seriously, the article is spot on. Storage has become nothing but more complex. While everything else seems to be getting simpler (dumb hardware and smart abstraction) storage seems to be going the other direction UNLESS you are big enough to write your own storage layer. CloudStack here I come!
I can't wait until the day I no longer need a dozen Cisco CCIE's. They are the hardest employees to manage, the ones that cause the most outages, the hardest to talk to, and the ones that seem the least to grasp the concept of dynamic infrastructure. They have held up the software defined datacenter too long. I am a big proponent of SDN, but not sure i'd stick with a Cisco driven model. I'm going to wait until the Gen2 gear arrives, then slowly replace everything with the cheapest gear (HP, Huwaei?) I can and leverage the move of the control plane into the software layer.
I don't care
I really don't care that states spy on each other and try to tap into other countries communications. I think that's just the real world, and is a requirement in keeping a nation secure.
What worries me is when these abililities filter down to local law enforcement agencies. I should be able to communicate and store infromation without worrying about big brother government spying on my for whatever reason tickles their political or financial fancy on a day to day basis. That's what is worrying - that recent terrorist incidents seem to have bred a "right" and desire for government to spy on their own constituents.
Who exposes the management interfaces of infrastructure on the user network? Surely every man and his dog restricts access to these interfaces to a management network which is ACL'd and monitored?
The Real Reason is...
It really is historical. Microsoft run what amounts to an internal exchange rate for each country. This applies both a real-world exchange factor, as well as allowing for adjustment to suit market conditions - for instance providing deep discounts in developing markets.
Many years ago, Microsoft set a rate of aroun 0.60 for the Australian market based on the currency at that time. Since then, thye have simply not adjusted it as the Aussie $ moved upwards in value. Why? Well to do so would massively affect the APJ budgets - quite simply the same amount of revenue would need to be made up for in another market to meet their quarterly / yearly numbers. Nobody wants to have to find a 40% increase in revenue as well as their yearly 10% quota increase, so it's simply in nobodies interest internally to make it happen
EMC ALWAYS talk Post Dedupe
Every EMC sales guy I have ever had the fortune of speaking to (and they knock on my door weekly) always talks values post dedupe. If you go and look at the specs for the datadomain devices for instance:
Up to 6.6 PB Logical Capacity
Up to 133 TB Usable Capacity
It only give you 133TB RAW, but the sales guys can tout a 6.6PB figure. Apply that scale of expected dedupe to the original problem and your 233 racks becomes 12.
NT4 also killed Novell. Remember the Gateway Services for Network? It allowed any windows PC to connect to your Novell Netware network and all appear as a single client. Microsoft didn't enforce CAL's, so you could buy one copy of NT4 and save a fortune in licence costs. And you could use the same key on all your servers... I knew many organisations that did this to save money, and this resulted in no more Netware purchases and a slow migration to Windows based services. Very smart from Microsoft.
Re: What do we need
While a nice competitive landscape sounds great in theory, in practice it has failed Australia miserably. Mostly because of the way the Libs sold off Telstra with a virtual monopoly over the pits and pipes which has prevented any kind of competition apart from the most dense areas. But the reality is that the last 15 years of competitive Telco landscape in Australia has resulted in a broadband backwater.
In that time, Telstra has also let the CAN decay. 10 Years ago, Telstra declared that the CAN had 10 years left in in MAX, and basically started a process of running it down to scrap value. We now have a copper network that is really worth only the weight of the copper - the pit and pipe infrastructure is the real gold-mine. Sure Telstra knows they have to spend $300mill fixing it, but that's pocket change for a country the size of Europe.
And what's this magical 400Mbit service that's available? Are you talking the HFC network that they stopped expanding 10 years ago and only covers less than 10% of Australia's houses - once again in the most dense areas already?
Many of the benefits of the NBN will be it's ubiquitous service. You can't plan on wholesale telehealth when people's upload speeds range between 1Mbit and 20. How do you do mass rollouts of VC solutions to the elderly when you have no idea if they will work or not? How do you enable the regionalisation of Australia's workplace when speeds and latencies are all over the shop? Under the Fibre NBN plan your physical location becomes almost irrelevant - not so under a variable speed copper plan. And Australia cricically needs a way to both boost national productivity while dealing with the massive infrastructure problems our capital cities already face.
Yurware is also a spin-off. AWS DR tools. They do some pretty cool research there.
It will be a shame really if they can't make enough money to keep going.
The NBNCo corporate network is also COMPLETELY separated from the operational network. Typical politically driven drivel from one of Murdochs papers.
I think you are actually reading far too much into this. I've spoken to highly placed EMC architects about what is coming, and it is nothing more than a unified management platform, but promptly cloud washed. There's no new real capabilities as such, just a way of managing all the various products that EMC sell you through a single management framework.
Flywheels can be scary
Personally i've never trusted a Flywheel UPS ever since I saw one go "bang' in a datacentre. It had been service 2 days earlier, so the suspicion is that the engineer did something wrong, but due to the large amount of damage (it basically destroyed itself) when it let go, nobody will every know. Oh it it's trail of destruction included cutting through some HV cables which required the complete shut-down of power to the facility.
Outlook on RT works fine
So I am running Outlook on the RT. The installer is not integrated into the store, it's just a normal setup.exe style solutin which runs a very normal looking office installer.
It appears to work just as well as Outlook 2013 on Win8. I haven't run into any power issues whatsoever, and the build date of the product is from November last year. It seems to me that it's a purely marketing decision not to release it, nothing technical.
I'm now slight concerned. Have just switched to Cloupia after losing half a year and countless dollars following Cisco's CIAC (Intelligent Automation for Cloud) down and endless hole of customisations to provide basic functionality, i have come to the realisation that Cisco just don't get software. They don't understand it, don't know how to support it, don't know what it takes to be a software company.
Cloupia in two hours gave us more capability than Cisco services could in six months of custom dev. It's sad when a third part can automate a Cisco switch but Cisco can't. Now I'm back having to deal with the TAC which might get hardware, but simply doesn't get software.
I also wonder if they will slow down the development of the HP / IBM / Brocade / Huwaei functionality.
I dont' see the fuss in building a completely sealed device. I have sensitive dive electronics in cases that stand pressure changes 10x more than you'd get taking a drive up to the top of Mt Everest.
What i'm not happy about is yet another industry using up Helium which is already so expensive it makes deep diving a seriously expensive hobby. If it goes up much more, it will spell the end of technical diving as a hobby :-(
I've used BT software quite a bit doing 10's of thousands of Notes to Exchange mailbox migrations. The software is a piece of **** really, and the support from BT is woeful. I've never met a worse company for support - they only support the absolute latest point version which means every time you hit a bug, you have to upgrade your migration platform, which invariably introduces yet another bug. And when you point out that they change functionality that you were reliant on, you are screwed as they refuse to support the old verison, and won't change functionality without you paying for it. Worse software company ever. Oh, and they only provide support 9-5 NY time.
72TB of storage in an enterprise managed cluster for $10k? Please tell me where. And i don't think you can assume a 10x multiple for bulk discount. That's a little ambitious.
I'd say either
a) someone has managed to infect their image source (noice exploit!)
b) the local reseller / PC shop has some idiot working for them that is infecting machines
I seriously doubt samsung would deliberately install a very easy to detect key logger!
They don't get the other side either. While they suck as a content provider, they also suck at the content delivery.
For years they have insisted on only providing content on their limited HFC network which only passes a fraction of the population. They tested IP STB's years ago but needed to recoup the massive cost of the HFC deployment where they decide to race Optus down the same streets.
They have done everything possible since then to try to stop the deployment of IPTV until recently deciding to deploy an xbox client.
What they should be doing is working WITH every major ISP to deploy foxtel via IP. There's only 5 major ISP's in australia - how hard would it have been to put a connection into each one of those ISPs cores and then worked with them to deploy multicasting.
Instead, they now have to play catchup with FetchTV who have played a far smarter game.
- Vid Hubble 'scope snaps 200,000-ton chunky crumble conundrum
- Updated + vids WHOA: Get a load of Asteroid DX110 JUST MISSING planet EARTH
- 10 years of Facebook Inside Facebook's engineering labs: Hardware heaven, HP hell – PICTURES
- Very fabric of space-time RIPPED apart in latest Hubble pic
- Massive new AIRSHIP to enter commercial service at British dirigible base