Cloud computing Engine: Automated technology for on-demand provisioning services

A day ago by the BMC Software Cloud Management technology conference held in Shanghai, the experts will be on the cloud computing vision, cloud computing applications, cloud management, Business Services Management (BSM) and other topics to launch a fascinating discussion. The following is an excellent speech by the BMC China Senior Software Advisor, Mr. Chui:





in fact, relative to the service portal or monitoring tools, automation in the cloud is relatively disadvantage, because the automation technology is implied in the cloud computing platform behind, to the customer that is not see the service interface, operators can not see the very beautiful monitoring interface. But just as everyone chooses a car, the quality of the car's engine often determines the value of the car. Automation is to cloud computing, just like the engine of a car. For the rest of the time, I'll be introducing how the BMC will provide a complete cloud computing service to customers through an automated solution in cloud computing.





First please look at two pictures, the upper left corner is the post office, in the past there is no automatic mail sorting machine need to spend a lot of manpower to do such a classification, so in a region there may be dozens of or even hundreds of people specifically to do the classification of mail, but when the 90 's introduction of this automatic mail sorting machines after the overall efficiency greatly improved, This is the benefit of automation. The second photo is the 80 's telephone program-controlled switchboard before the introduction, we call need to have a dedicated personnel wiring, a person can only handle more than 10 lines; but with the telephone switchboard, today a communications manager has been able to charge tens of thousands of users line. So we can find that even in the traditional industry, after the adoption of automation, the first is the overall cost of decline, because the reduction in manpower. Second, efficiency has improved. Now the speed of mail transmission, and even the speed of our call, with the original is not the same. In addition, for us as a user is a service experience changes and service quality improvement. These two are very good examples, telling us that even without it, there is no cloud computing, and the traditional industry also needs to have automation technology.





Today we talk about cloud computing, for a business, we think is actually a tortuous road, which will experience the following four stages, we call it the ability of the cloud. First, the ability to have a single point of equipment. First I need a basic architecture to build a virtualized environment on this infrastructure. Second, the infrastructure needs to be automated and able to drive the device in an automated manner, possibly a virtualized platform or even a component-level object such as a database in a virtualized platform or middleware. Third, the integration of management capabilities, we need not only a single point of equipment processing capacity, but also hope that the cloud platform to provide end-to-end management capabilities. Ultimately, we're going to be able to capture the user's needs from a service perspective, with the aim of enabling IT resources to be delivered directly to end users in ways that are understandable to the user. These four capabilities are needed, whether private, mixed or public.





in 08 when cloud computing first started in the industry, the BMC found that the idea of cloud computing was basically consistent with the concept of BSM, because cloud computing was equivalent to a best practice for BSM (business Service Management). Automation is a very important part of this best practice, from application automation, database automation, server automation, and network automation to the capabilities that cloud computing requires. Some people may ask why we need it automation, manage it today, and manage it well if you don't need it, and it can cost more to automate it. We explain why it is needed for automation in the following four ways. First from cost considerations, a server management cost is basically equal to the cost of buying a new physical server three times times. Today we have to virtual multiple servers on the physical server, so in fact, we are facing more management objects than the original, then the increase in management costs is no doubt. Second, quality. According to each tripartite agency survey, 80% of all IT failures are due to improper configuration changes. In this case, we refer to it automation to streamline and rationalize the configuration steps to minimize human error. Third, 90% of the problems are known and avoidable, in the IT automation category we need to do some compliance checks to be able to before the problem, through the means of compliance inspection early detection of some of the technical risks and vulnerabilities. Finally, the issue of applying the publishing speed. Today, no matter which industry, Enterprise's business system is more and more complex, involved in the logical components and related parts will be more and more, for enterprises to apply the link required for publishing complex, usually the application of the release time than expected to exceed 60%. These times cannot be shortened without the assistance of automation software.





Based on our BMC experience in many automation projects, we have concluded that there may be four phases in the process of moving toward automation, namely, standardization, scripting, product and service. Standardization means that at this stage, the enterprise may realize that I need some process of IT operations, although I do not have some automated tools, but I can through the people, through the documentation of the operation of the IT calendar to solidify to form a standard. We use this standard to perform operations later when it comes to the same kind of operation. The second stage is scripted, and when I have standardized, some of the simple, standardized IT operations processes that were set up in the past can be scripted, allowing internal IT staff to write scripts, send people to run some scripts on a regular basis, or use crontab to run scripts automatically. In the third phase, when the script is used more and more, the enterprise will take into account that I want to refer to some products, may be for the server automation, may be for the desktop, may be for network automation. The fourth phase is service. Service more refers to the concept of automation in cloud computing, where automation is not just for the IT operational sector, but for automating the delivery of IT resources to end users, which we call the concept of service. For most businesses, it's not necessarily going through these four stages, but it's basically going to go through these things, maybe three stages, maybe two, but you need to do these things.





The first stage we call the standardization phase, which of the things we can standardize the process of it? Our customers in the bank is more common, is to do a daily patrol. Early in the future to arrange for a person to log into each server, to knock a command or multiple instructions to view the state of the system, or sometimes do not do normal monitoring, may want to look at the application of the system configuration file, these are the day-to-day operation. There is also an example, we often have some business system upgrades, generally speaking, a set of fixed business system, I upgrade the steps are basically fixed, from doing database field table modification, to the application of the file distribution, or file decompression and so on these are standardized processes. Companies will solidify these things as it processes to form a document and hand them over to the people below. First of all, regardless of other circumstances, regardless of human costs, regardless of the error situation, we think that this is better than no process at all. But we can count the workload. For example, today there are 200 servers may be a medium-sized enterprise needs to be managed, as we do daily patrol as an example, a person needs to log into a server to view the configuration file, log into a machine takes two minutes, 200 servers spend 6.7 hours a day, if a person to do such things, a weekly need to take 33.5 people, or 3.5 days, each year need 182.5 days. It's just a check, and we can often see that the customer's patrol list is often hundreds. When my patrol range is more, the more we consume.





so in the second phase, we can see if the problem is probably scripted. Here is a very basic scripting code, the script to the machine after the run to collect a data, and then output to the inspection results, looks like a script can reduce the time cost, because I just send this script, and finally recycled a feedback value, I can complete the work, The big deal is I'm going to take another sign of this value. But what's wrong with scripting? First, how the script should be directed to publish, I just talked about a generic script, but most of the time, our server group is based on my business system type division, I need to check these items is not to ensure that every server is the same. The second question, what to do with this check request, we also need to send someone specifically to modify the specific script. So we have three major problems with scripting, and the first category is unsafe. The script appears in plaintext, including a username/password if you need to log in, plus additional access information. The second is difficult to maintain, when you make changes, how do you maintain the script. The third part is difficult to manage.





So in the third stage, the enterprise consider whether I can quote some industry mature products. In this diagram we take the example of application publishing, basically, we will involve four teams, first of all the business system will have the application development team to carry out an application system packaging, after the package in the real business package deployed to the production environment need to do some validation, this is done by our testing or QA department, Next, when these packages have been sent down, the operations team will maintain the maintenance, in addition to maintaining the business system, but also to ensure that your existing server system itself above the operating system version can be upgraded according to your application requirements. Part IV Security management teams may also need to be checked regularly. Generally speaking, if it is a product, we will refer to different products. But the BMC provides a complete set of automated management platforms that can be used to fully implement all of the aspects I've just talked about, from OS to database to middleware to application, from all aspects of publishing to control.




At the
automation level we will involve automation from the underlying network equipment; second, server automation, whether it's a physical server, a virtualization platform, a third is database automation, and middleware automation and application automation. Next to do a comparison, we do a business system on-line, the following is the traditional manual way to do, above through the BMC automation software to complete. First you can see that in the manual phase, may be by a business system users to make such a change request, a new business system online we need to buy the server, even if not to buy a physical server today will need to deploy a virtual server, so I need to have a special person to do. Next to the network management, in accordance with the requirements of the enterprise itself, the designated IP binding, my server access to the network. Next, the server to put into production environment, need to play patches, need to have specialized technical personnel to do server reinforcement. The first three parts are done after the deployment of the business system, dedicated to the deployment of a business system application. Finally, the enterprise considers that this system may be oriented to a larger number of users, but also need to add a load balance. In a real enterprise environment, of course not every environment the same person, perhaps the same person to do different things, but the whole process time is relatively long, from the bottom of the base to the application. Through the BMC automation, we can achieve: first, the user proposed such a change request, automated management software can be operated through the logic of the operation of the process to link it up, when this step fails, what should be done, through such a set of platforms so that the business system can be quickly online. Of course here, we have not yet to the cloud, which is in the stage of automation management.





The fourth phase, the cloud computing phase, is called service, and automation is service oriented. Automatic management of the corresponding is resource management, first of all, resource management is divided into several large chunks, one is the server and application automation, and the other network above we can directly to the physical network equipment to do configuration. Another is the automation of the database, in the cloud environment to deploy the corresponding database software. Including for storage, it can be partitioned directly to the level of storage physics. The BMC Cloud Lifecycle management can support the industry's mainstream platforms and systems, such as VMware, Xen, and other mainstream virtualization platforms. For the web, we open boxes to support Cisco networking equipment. We support NetApp in storage. These are the support of existing mainstream products. What will we do in the future? We can see that we have a provider API here in resource management, which is an interface that the BMC leaves to more vendors in the future. If you have a new device for your customer, you can use the API to support it, and through the API you can call a Third-party device's professional management platform for unified management. Second, the Cloud Platform Administrator service Blueprint, which translates service definitions into real it deployments. In the deployment, I can be deployed on a single VM or two VMs, in addition to installing the operating system in the VM, the next VM is to access to the network level, we will be at the network level through the virtual network card configuration, which do things far less than we see so simple, It does a lot of automated operation configuration underneath. Third, end-to-end automated deployment. In our program, we carry out operating system deployment and application deployment, and provide operating system reinforcement. What is reinforcement? When you're done with these applications, you need to do some compliance checks, which are done during deployment, so we call it end-to-end, not just on a level. The fourth feature is that we can configure the resources according to the service level. For example, in a manufacturing industry, today to use cloud computing, I may have to put two business systems into the cloud, one is the OA system, one is for the outside user's website. Our policy engine is the technology of labeling, which chooses the level of a service before the business system is deployed, based on the service level to determine which resource pool the business system should put in a predefined service level, and then invoke the Resource management module to complete the real deployment.





Our network automation configuration can be divided according to many tenants, in the network deployment will do VLAN automatic division, to ensure that different tenants of the data flow in different VLANs. In addition, CLM Automation network deployment, not only configure the router switch, but also support such as firewalls, load balancing equipment. This is what we call the security of multi-tenant user data at the physical level. The second security related is that when my business system goes into the cloud platform, it needs regular compliance checks, one is the compliance of system-level patches, and the business system itself requires a compliance check as required by the company system. At this time, the automation requirements for the cloud management platform are not just software or operating system deployment, but the ability to check compliance. Therefore, from these two aspects to ensure the security of the user cloud computing platform. One is the application level and the other is the network physical level.





next two cases, the first case is the case of service automation, the customer is Morgan Stanley. In terms of business requirements, since Morgan Stanley has two service centers, they found that in the process of upgrading the business system, first of all, it takes a long time to do the labor, no way to ensure the continuity of the business; second, doing manual work sometimes causes unplanned downtime. So in this case, they went through a multi-party comparison, finally chose the BMC BladeLogic solution, through the bladelogic to help them improve staff efficiency, estimated to save 270,000 dollars a year. The second case is the case of the public cloud, the customer is Australia Telecom, the use of CLM, the day-to-day operation of Australian telecom and installation tasks spent time from day to minute level.





Each enterprise's it vision is different, you are now in the automation of the first stage or the second phase, or hope to the third stage, the fourth phase of progress. BMC as the industry's leading manufacturer of automation, we hope that through a wealth of solutions and projects in a large number of rich experience, can provide more support for everyone to achieve IT management goals to make our best efforts, thank you.

(Responsible editor: The good of the Legacy)