Cloud is increasingly becoming a familiar topic for many companies. Although it is still early in the lifecycle, Cloud Computing goes Mainstream, according to the BBC. And I am seeing an increasing amount of large enterprises discussing the subject with ever increasing understanding. Many are testing the public cloud, but a number of issues, including security and legal ones, are keeping them from using it effectively for their key requirements.
The public cloud ecosystem consists in three key parts, the cloud infrastructure (IaaS), the network and the devices. Although there are these security and legal aspects to take into account, the cloud infrastructure is becoming more robust by the day. As times passes, new services with higher levels of security, more business tailored service level agreements, and more business oriented services are starting to appear on the market.
The device space is still in its infancy. It is increasingly becoming clear that the PC or the laptop will no longer be THE interface into the cloud. As the Millennial Generation, also referred too as Generation Y, is entering the job market, they bring with them a complete new approach to interacting with technology. They use smart phones, tablets, iPAD's or slates etc. to surf the internet, to remain in contact with their peers and to interact with each other. They expect a similar approach in their business. While PC's and laptops will still be sold in large quantities over the coming years, building new devices allowing such interactions with the cloud will be the new battlefield. Highly visible battles between companies such as Adobe and Apple are the first examples of what is doomed to happen. Indeed providers of cloud services are looking at a single standard, being de-facto, to develop the interfaces allowing them to offer their cloud based services on any device, going from a set-top box to a smart phone. Many see Adobe Flash as just providing that capability. This however works for everything except the Apple camp.
Who will win this battle is uncertain at this point in time, but what is interesting is that the battle is happening and that many are tipping in. People are looking for an intuitive experience to access the cloud, regardless of the device they have chosen to use. It is actually interesting to see that a number of enterprises are no longer dictating which device their employees should use, but rather give them a compensation to buy the device they wish. Compatibility in the device world is quickly becoming an issue that needs to be addressed.
But the device side is not the only one. There is also the network. And here things are more complex. First, networking is quickly becoming the most expensive cost element of your cloud bill. First, the cloud service providers charge you per GB of information you upload or download. And if you do anything else than a compute intensive task, these start counting pretty quickly. Second your network provider also charges you for transferring data over the internet. So, you are being charged twice. But the cost is not the only aspect. There is also the real bandwidth available for uploading and downloading information. I have checked bandwidth during my travels over and over again, and found download speed going from 3-7Mbits/second, and uploads from 300KBPS to 1.4MPS. Even with the lather, transferring 1GB takes more than 15 minutes. You may argue that you have way better numbers in your premises, and I believe you, but the whole objective of the cloud is ubiquitous access, this means access from any place. Indeed, the cloud facilitates the work of the mobile worker, giving him/her access to his/her computing environment from anywhere in the world. Unfortunately, then you need to take consumer type network bandwidth into account.
It is the combined experience of the cloud services, the network and the devices that will make cloud really attractive to the mobile worker of the future. Today I see activity happening in each of the areas, but unfortunately, I do not see a lot of work going on where all three are looked at in combination. Maybe this is a new area we need to focus on moving forward.
I have been in India these last couple days and had the opportunity to speak at an Automotive event in Chennai. One of the presentations there intrigued me as the speaker bluntly pointed out that "Manufacturing is moving East". I went digging a little and found an interesting document from Eurostat, titled "Industrial Product Indices - Global Developments" . Unfortunately, the one I found only goes to 2007, pre-dating the latest recession. However, the numbers are staggering. Taking an index of 100 in 2000, China is at 240 and India at 160 in 2007. Contrasting with that, the 27 countries of the EU are at a meager 112, while the US is at 110. You could argue that moving east is maybe not completely correct as Brazil is at 125, but still far away from the India and China numbers.
The internal market in China grows as demonstrated by a growth of 17.2% of retail and consumer sales during this year's lunar new-year as reported by the Global Times. This has a number of implications.
First, as the internal market grows, American and European companies seeking growth are attracted by the Indian and Chinese markets and increasingly invest in the local markets to have/increase presence in the market. This brings an interesting new competition to the local companies. On the one hand, the Europeans and Americans have to learn to do business in Asia, but on the other, they bring efficiencies and production methods with them that local companies not always used.
Both China and India have experienced the American and European production methods through the outsourcing of manufacturing and services. Many Indian and Chinese companies are subcontractors and suppliers to large European and American enterprises. They have had the exposure allowing them to compete effectively in their local markets. However, smaller companies may have more difficulty and may end-up either going bankrupt or being absorbed by larger ones as they struggle to survive.
But as one of the attendees pointed out to me, globalization is increasing fast. He fully agreed that manufacturing was moving east, but pointed out at the same time that a number of Indian and Chinese companies had started buying European and American companies lately, citing Tata Steel buying British Steel, Mittal Arcelor, Tata Motors Jaguar-LandRover, Jelly Volvo, and I'm sure there are a number of others. So, in the end is it "Manufacturing moving East, Investments moving West"?
I don't know, but it definitely demonstrates increased globalization. And this brings with it a number of challenges for all parties. First, companies increasingly require working in a multi-cultural environment. This often, and I can speak from experience, brings with it misunderstanding and frustration. Indeed, the same word does not always have the same meaning, and decisions are not taken in a similar way. Doing some research for this post, I ran into a quite interesting article titled "Cross-cultural differences in decision-making styles: A study of college students in five countries", pointing to such differences. Now, such decision making differences do not only affect business people involved in the supply chain, but also the consumers. A study performed in Columbus, Ohio, researched the difference in decision process when buying a car between Japanese and American students. It's often difficult for management, located in one country, to fully assess the differences in the market in another country. They need to rely on local people and trust them enough to involve them in the decision process.
Second, as supply lines become longer and more complex, companies need increased communication levels to keep abreast of the situation. History has demonstrated at multiple occasions that it is often supply lines that stop the greatest worriers on their way to conquer new territory. The same applies for businesses. Building strong communication and supply lines is critical to embark on this movement east.
Local companies on the other hand have to increase their efficiencies if they want to maintain their competitive advantage, the knowledge of the terrain.
So, the current trend to globalization, the move east are great opportunities to develop more robust enterprises that take full advantage of their size and capabilities to conquer new markets, east today, maybe south in the near future. Whether these enterprises are from the west or the east, does not matter, it is the ones that can combine a strategic vision with strong operations and a deep understanding of cultural differences and global collaboration, that will be the winners in the future.
When discussing Cloud with enterprise CIO's in Manufacturing, the question on what applications to migrate to the cloud, and I mean here in particular the public cloud, often comes up. Many cloud providers tell them that the cloud is the future and everything should be migrated. Many of them are doubtful, and I have a tendency to agree with them.
So, the question remains. What applications should we migrate to a public cloud? First, lets come back to a point I made in an earlier entry, titled "Cloud Computing, Evolution or Revolution". The cloud brings a true differentiation when applications use multitenancy programming principles. This approach comes from Web 2.0, and allows millions of users to share the same code base, making the serving of large user communities much more efficient. The multitenancy concepts are not well documented but include design for failure, redundancy and fault tolerance. (James Hamilton - 2007)
Most enterprises do not have multitenant software deployed in their IT environments, most probably don't need such software for the functionality running currently in their datacenters. And most are not prepared to redevelop their applications in a multitenant approach. The majority of the large software providers are not prepared to do that either. So, using multitenancy as an argument to transfer applications in the cloud is probably not a good idea.
The other reason enterprises may want to move to the public cloud is the financial aspect. At $0.08 per CPU per hour, this really sounds cheap. Unfortunately, that is only a very partial view of the world, and when you start calculating the cost of server, operating system, memory, storage, I/O bandwidth (in and out), public IP address etc. the cost goes up very fast. Adam Goldstein developed a useful decision tree on determining the true cost of hosting servers in the cloud.
With that in mind, the best use of public cloud services is probably related to addressing peak capacity required at specific periods in the year. Using cloud services for occasional, short periods of time makes sense, using them continuously may end up costing more than hosting the service in house.
So, for manufacturing companies, what does that leaves us with. I would believe a couple categories of functionality could be hosted in the cloud:
- Workloads addressing new requirements, for which no solution is implemented in the datacenter. I would use as example, supply chain collaboration applications, collaborative product development tools.
- Workloads related to marketing or after sales services, where the company expects large amounts of users. As the applications still have to be developed, it would be mandatory to develop those using multitenant architectural principles.
- Occasional needs for compute power, for example, simulation and testing software for product development. This is often referred to as "cloud bursting", but frankly it is still in its infancy.
In each of the above, as Adam Goldstein points out in his decision tree, the first element that needs to be looked at is the sensitivity of the data that will be used in this cloud hosted application. This brings us back to the security aspects we have already discussed previously.
My remarks are focused on the public cloud. Obviously, using modularization, virtualization, standardization etc. companies may want to transform their existing datacenter in a private cloud, as long as they are able to run their applications in a virtual environment. Doing this will free up compute capacity that could be used for some of those new applications, and avoid having to move to the public cloud.
In a nutshell, the public cloud should be looked at for new applications and for occasional need of extra compute power. Existing datacenters should slowly but surely evolve to private clouds in which modernized versions of today's applications run. In the future both may merge, but that is still some time away. I remember 30 years ago, the arrival of the minicomputer was going to be the end of the mainframe, well some are still there, isn't it. I'm sure the same will happen with today's datacenters. Large corporations are not ready, and have no real incentive, to move everything to the public cloud, despite the wishes of the cloud advocates. And you believed cloud was something simple?
There is one process we know very well after one year of recession, and that's cost reduction. Most companies have scrutinized their costs and tried to take every waste out of the system. Many were quick to point out that, although the sustainability agenda was still on the radar screen, it has been taking a back seat for quite a while now. I cannot judge whether it has or not, but would like to argue it should not have. Becoming greener is an excellent way to reduce cost if you look at it in the right way. In this entry, I want to take a minute to explain you what I mean.
It all started about one year ago, when I heard a radio interview from Jean-Pascal van Ypersele Vice-Chair of the Intergovernmental Panel on Climate Change (IPCC). He pointed out something extremely simple: "The best energy is the energy that is not used". Just think about this. There is a lot of focus on using green energy, and I do not mean that to be wrong at all, but what he says is that before turning to green energy, take a hard look at the energy you consume and see how you can reduce your consumption. Whether green or not, consumed energy costs money, avoiding consuming it helps you reduce cost, isn't it?
Well, that is exactly the point I want to make. Let's now look at our Supply Chain. We consume energy in a variety of forms throughout the ecosystem. Electricity, gas, oil, petrol and you name it. How can we get a consistent view of the energy we consume? All above fuels have one thing in common: they generate CO2. You can argue that electricity may come from non-fossil sources and as such do not generate greenhouse gasses. Let's put that thought aside for a minute.
If I can develop a CO2 emission profile across my supply chain, I get to understand very quickly where energy is consumed. I can now focus on reducing the consumption at that step in the process. Doing this may imply I have to change the design of the product, use a new manufacturing approach or change something somewhere in the supply chain. It may happen that a slight increase in energy consumption in one place may result in large savings somewhere else. It is important to include design and engineering in the process to ensure we go back to the root cause of why so much energy is consumed. Root cause analysis is nothing new; we have been doing this for quite a while.
But how do we gain that CO2 view of our supply chain. Quite a while ago, I wrote an entry focused on two new acronyms, BOC and BOS. BOC, bill of carbon, gives the answer to our question here. What do we need to calculate the BOC? Fundamentally two things, first the complete bill of material of the end product, and second the amount of CO2 generated at each step in the manufacturing process for one unit of product/component/ingredient. The latter is mainly based on the full bill of process. Each partner in the supply chain should calculate all emissions related to the operations under his control. This includes manufacturing, transportation, warehousing etc. Whether those activities are performed by him or by a subcontractor, do not matter.
As stated elsewhere in this blog, gaining such information from suppliers requires the existing of a trust relationship between the partners. Once we have the information, we can start the review, leading to improving the energy consumption. However, to maintain the trust relationship, we should never forget to ensure the partner also benefits from the exercise, otherwise he will not be motivated to continue collaborate in the future.
Now, let me come back on the use of non-fossil energy. From a cost reduction perspective, there is no difference where the energy comes from, but from a sustainability perspective, there obviously is a huge difference. However, I would dare to argue that reducing the consumption all together, leaves the non-fossil energy for other companies, reducing the CO2 emissions in the same way.
Looking at CO2 is a great way to analyse the energy consumption throughout the supply chain, looking for wasted energy and related costs. It is an interesting way to reduce costs, improving the bottom line. What are we waiting for?
Last Wednesday Toyota temporarily halted sales of 8 models in a recall related to a problem with gas pedals. This was enough for Industry Week and the Agence France Presse to ask a fundamental question: "Has the company sacrificed its legendary quality to become the world's number one producer?" This recall that may affect more than 2M vehicles and is being extended to Europe and China. Actually this issue was first raised last September on the Toyota USA newsroom as a consumer safety advisory. A remedy was described in November 25th. Quickly scanning the Toyota Owners Club forum, I found indications of a pedal catching the carpet as far as June 22nd, 2009. And I only scanned a couple minutes.
My objective is absolutely not to want to blame Toyota for what they are doing. My point is different. With the proliferation of the Internet, a lot of information regarding the experience gained using a product is available in blogs, forums, on twitter, facebook and a variety of other social media. Are we looking at that and do we take notice.
A little after the HP/Compaq merger, the warranty of one of our notebooks went through the roof. Our product engineers had no clue of what was happening, and, with the objective of solving the problem, they tried to analyse the repair logs. Not having any particular tool available (a lot of those logs consist in notes typed by repair engineers), they loaded hundred of thousand of records in Microsoft Excel. It took them about 6 weeks to notice that, every time the modem card was replaced, the same happened with the keyboard. They could not see the correlation between both. They knew the modem card was a small printed circuit sitting above the motherboard. They also knew that, to access the motherboard, the keyboard had to be removed. But that's about the only link between both. So, they went to talk to the repair engineers (mostly done by third parties) and discovered the way to remove the keyboard had not been well described, resulting in many of them pushing a screwdriver under the keyboard and lifting. I let you imagine the result. We had gotten a bad batch of modem cards, but our poor documentation made things worse.
We could not allow this to happen a second time. This is why we put an early warning system in place. We upload repair logs, and a lot of information from the web in a system that analyses the unstructured data and highlights recurring elements. This allows us to very quickly spot potential problem areas and check whether those require a recall, a redesign or both.
In working closely with our distributor and repair channel, we receive the raw information required to perform the analysis. Working closely with suppliers and our own product development & engineering, we address issues early. In doing this we have saved millions of dollars over the last 6 years, being able to limit recalls to much smaller batches, which is obviously way less costly.
Maybe Toyota could have done the same, spotting the problem earlier, limiting the amount of cars they need to recall. At the same time a couple accidents might have been avoided. Toyota is not alone in this situation. This is common throughout the automotive and electronics industry. This is why we developed a customer version of our internal system to help the industry reduce warranty costs and spot issues early.