May 23-May 25, 2012 cloud base of China Cloud Computing Conference opened the curtain, show cloud practice, explore cloud trend, share cloud future. The cloud base shows the latest technologies and new products in the field of cloud computing, and discusses the development trend of cloud computing, the great value of large data, the energy-saving of data center in cloud times and so on. Tian Yun Technology product manager Guo Hong for guests to introduce Skyform cloud platform.
Everybody good afternoon, Sky Cloud technology in May officially released the Cloud Platform 1.1 version, now the name of the product is called Skyform. Let me introduce you to the main features of version 1.1. First of all, we divided into several modules, just now Zhang also introduced, our cloud platform is the concept of IT service, we emphasize the service, which you can see, we in the 1.1 version of the main services provided.
Cloud Technology Product manager Guo Hong part of the network services, the most basic is the virtual machine services, which is based on virtualization, but also our early 1.0 version of the Cloud platform. Our new version adds some network services, such as virtual load balancing, virtual firewalls and IP. minicomputer services We are also released, we can support the minicomputer partition, while we are on the basis of X86 virtualization we have added some like monitoring services and backup services, and so on.
overall thinking, the entire realization of IT service. The overall development trend mentioned by Zhang just now, In fact, our 1.1 version is still positioned on the basis of SaaS, our service is each individual component, the user wants to put these components together, constitute its application, this is a typical SaaS foundation, we have SaaS service basically finished, PAAs will be next.
The
1.1 release is divided into two modules, a resource management module and 4 modules. Operation Management is mainly to the whole lifecycle management of these services, resource management is mainly to the physical resources, at the same time supporting a variety of heterogeneous resources. In the resource pool This module, we are divided into 4 separate modules, this module is sold separately, our Resource Management Module X86 module is the basic module, the remaining modules are optional, will be based on your business needs to choose.
Services we divided into several categories, first in the X86 virtual resource module under the basic provision of these four kinds of services, virtual machine service is the entire application lifecycle management of virtual machines, fast storage We also called EBS, I can apply for a single hard drive, can be considered a floating hard drive, this hard disk can be hung on a different virtual machine. Virtual machine backup service can be separate from the virtual machine services on the basis of the virtual machine backup, its basic idea is to take snapshots, backed up through the snapshot, you can restore the virtual machine by backup. The monitoring service is for the entire virtual machine service, it uses the historical situation to be counted, has the report, you may inquire your history. For the network section, this is the most important feature of our 1.1 release. You know, before this, if you want to butt with your network equipment, all the network is basically manual, now the hardware device slowly support the virtualization function. So, like firewalls, load balancers This is typical in the Web application of a large number of needs, now the traditional application is each set of applications you have to bring a set of hardware, so that in the cloud time physics machine can become a virtual machine, in the network device is not also can be virtualized, so, now hardware manufacturers are slowly supporting the virtual network equipment , can make a physical network device into a lot of virtual, each user you apply for a virtual later, your use and you use the physics is exactly the same, we are integrated with the hardware equipment, the current version of our support H3C hardware equipment, other equipment we can do two times development. At the same time we support the public network IP Services, public IP services are mainly able to apply for a public network IP, bind the public IP to your virtual machine or your network device, so you can access its virtual machine or its application through an extranet public network in this way for applications that use public services.
minicomputer Service, currently this version is mainly to support the Solaris Container, is a virtual operating system on the operating system, equivalent to a two-level operating system, it uses the same as our virtual machines, we are doing the management of minicomputer partitions, Let the user from the service point of view can apply for a small machine partition, as you apply for a virtual machine full user experience, many applications run on a minicomputer partition can be like to get a physical machine to use.
Let me give you a detailed description of the functions of several modules. For operation Management, it is mainly the management of the whole lifecycle of service. So, we introduced a concept called service template, this service template is a standardized format for end-users to choose from, and we currently have a service template that we basically support for fixed parameters, like you have different specs, we've got all the standard, like you're buying an Apple Computer now, Different models, each model is the standard configuration, we do not do the memory hard disk can float, the future PAAs can be considered, SaaS is standard, the entire service template has its own lifecycle, in principle, by the administrator to define, after the definition to be audited. The audit passed to release. After publishing, the end user will be able to look at the contents of the service template in your service catalog.
we've just talked about it. The focus is on IT service, to the user, it is to buy an IT resource or apply for an IT resource as you buy a book, the whole experience is the same. So, we use the concept of service catalog, can be all he wants to buy, all the items can be listed on the top. In this way, after the administrator through the service template released, through the service catalog release, so that end users can see what he can buy. Why do you need both of these processes, one for service templates and one for service definition? This is mainly from the management, according to our experience, operation management to go through such a process, the process requires two level of approval. The first level of approval is to verify that I do the template itself is the right, it released after the release, you can do internal testing, testing passed, I feel that this service can not open the virtual machine, or can be out of network equipment to do a test. After the test is passed, he publishes it through the service catalog, which the end-user can buy.
the second part is the management of a service instance, once my end user orders a service on my service catalog, the service becomes a service instance, and the cloud Platform system is responsible for your entire lifecycle because it has been opened and I want to ensure the quality of the service. So we're going to create the cycle of the entire service instance, followed by a look, and you can manipulate it. If it is a virtual machine can be turned on, you can turn off the computer. In addition, you can make some changes, if you need to change your service, and so on. But the end of the service instance is canceled, your time is up, or you finally don't want to use, will automatically recycle, add into the resource pool.
Order management, the whole ordering process is the same as we buy things, so we have a concept of shopping cart, a shopping cart can buy a variety of things at the same time, everything is independent. We have the entire order lifecycle management, including order applications, approvals and inquiries, and some expiration reminders. An order can be renewed and unsubscribe, and so on. In addition, as operational management, set up a cloud platform of the entire user system, the current user management we are divided into two parts, for the cloud, we support the user registration, for the private cloud, in principle, is the administrator to create users, not to refer to the user registration, because all users must be within the scope of his control. In addition, there are some user status management and password modification and so on. At the same time, we also support user groups and user permissions, what the user's role can do.
for the whole operation management, resource metering, we want to integrate with the billing system, the private cloud in principle does not need, also need to see your service usage, it is possible that the resources you have not used, or the entire resources are basically not boot or what. For the public cloud, we can implement billing management, with the billing system integration.
end users can see their own services through management, the entire process can be user-service, but also provide a service console, you can operate on your resources, such as your virtual machine for remote access, or the public IP mount, and so on.
in terms of resource management, X86 resource pool is mainly the life cycle of the virtual machine management, as an administrator we can add equipment, the entire equipment management This is the basic, for the life cycle of the virtual machine is mainly some operations with some modifications and so on, at the same time, we support fast resources, for fast storage, You can mount the virtual machine you need or uninstall the fast storage. At the same time, job scheduling. Scheduling, in fact, basically there are two kinds of statements, the simplest one is to fill a first, the first fill in after the second, so that the utilization rate increased, behind the empty machine can be turned off, and then open up all can. There is also a kind of rotation, each machine is allocated one, so that the user's virtual machine as much as possible to use the system's resources, because such resources are relatively rich, while supporting user-defined algorithms, can be added, such as a request for a CPU load the lowest machine, to dispatch me to this machine and so on.
at the same time, this version of the increase in resource monitoring, this part we do a lot of work, including the virtual machine report, added some historical data reports and resource monitoring alarms and so on, resource management is a unified portal, will all kinds of resources, minicomputer, X86 under the same portal.
in version 1.1, we provide a physical machine management module, currently still used by the system administrator, currently does not provide the final service of the physical machine. The main function is to be able to deploy the operating system in the X86 physical Machine resource pool, which can be deployed and deployed with standard operating systems.
In addition, we provide monitoring and alerting of physical machines, now in the rack or blade of the server basically have their own alarm system, we can the physical machine alarm, because their alarms are based on stand-alone, we will these alarm systems integrated into the cloud platform, into a unified alarm, the physical hardware of the alarm gathered in. At the same time the physical machine is deployed, the physical machine operating system we can collect information through SMP, we have it to implement alarm and statistical reports.
Network Management module has just been introduced, at present, we are based on H3C devices to achieve the firewall, load balancing, NAT and other virtual network functions, the current three functions are completely independent, so, from the system management, to define the three templates, to define specifications, end users in the application of resources, You can use these devices. For example, load balancer, you can specify which virtual machines of your user can be mounted to load balance through the User Service Portal. For the firewall, you can define the rules of the firewall, your application only open 8080 fracture, you can define the firewall and your application, only open 8080.
public network IP is to allow your internal network resources to connect with the external public network, you can open your application to your end users. minicomputer, the entire resource pool lifecycle management, can add the entire minicomputer physical machine, add to the resource pool, we can monitor the existing minicomputer, there are some running I container we can take over, while creating new minicomputer partitions, while on the whole of minicomputer resources, Lifecycle management, including its creation operations, and so on.
This is a typical case of the application system with IaaS, everyone in the traditional Web application, typical will have a public network, through the public network to come in there is a firewall, which is to protect its business, each application may have its own firewall. There is also a load balancer, and finally a database. Here the IP address is casually written, they may be different IP gateways.
For business, the business of the public network IP map to the firewall, this set if it is a set of application system, it to his public network users through the public network to visit. In fact, his Web server is a lot of platform, to achieve its load balance, many users in parallel access, but at the same time this system to do some debugging, how to solve this problem? Apply for another public network IP on the fire wall bound on the fire wall words may be equipped with only 80, 80, other fracture if you want to do not come in, the firewall is for the end user, in addition to use the public network IP binding virtual machine, you can the virtual machine for your debugging and modification. If you find this OK, you can put this dynamic binding to this machine, this is a typical public IP service to do things, with my internal virtual machine or network resources are good, to bundle.
This is the process of some of our services, not necessarily detailed introduction. The general idea is this, everyone through the service Portal can choose your product or service, and then into the shopping cart, into an order, after the order you can see the approval process, can be three level approval, which level has not passed, and we have a unit in the reimbursement process is exactly the same. After the passage, can be split into work orders, open, through the remote console or operation of your virtual machine operation, complete the approval, follow-up work can do mount, fast storage, or mount a public network IP, or define firewall rules or load balance, let it use the service through the network.
The
Fast Storage service is the reverse, it can be virtual machine, can hang on the virtual machine, but at present we only support hanging to the virtual machine, fast itself format to the user to do. Public network IP just introduced the entire process, it is characterized by binding, can be bound to the virtual machine, can also be bound firewall or load balancer, see the structure of your network. At the same time, you can dynamically change, you can bind later, unbind, and then do another. Public network resources This is actually very useful, because the public network IP is very small, very expensive. Therefore, the general user will not apply for more, you in the case of the cloud how to access your internal machine, public network IP is a very useful tool. We've all mentioned that this is very useful in the public cloud project.
load Balancing just mentioned, general Web applications need, here to mention, a physical load balancer can open how many virtual, from the current situation, the more high-end equipment may support more, because the memory needs of the larger, generally as long as you 20, Look at the specific model of the configuration. At present, we are a full range of h3c, we support, see the specific configuration, see how many applications you want to support.
Firewall is mainly a rule, each person can define their own independent rules, it's open this rule you can according to the IP address, according to the port, this is the comparison of some of the standard firewall content. Some advanced features, we currently support common firewall rules.
This is our entire Operation Management Service catalog example, this is the user from the service, everybody goes shopping, the entire experience and the shopping, this is our resource management platform, mainly is the monitoring resources management.
from SkyForm1.1, we provide NO. 400 technical support of the telephone service, currently mainly for our sales or have used our products, the phone is 24 hours of front-line support, we will have a second line support team behind.
finally to tell you Rodamap, now we release the 1.1 version, mainly network services, operations management and monitoring, physical machine management, etc., at the end of July, early August, we will release our 1.2 version, it is mainly to provide storage services and network services, some enhancements. such as network public network traffic, etc., at the same time, we have started our 2.0 design, 2.0 focus is just Zhang always mentioned the instance PAAs it's design ideas like Web application of the whole thing, if s build to build their own, installed on the above your application, responsible for the network connection, If as a whole instance, I will be able to apply the entire deployment, it needs all of the IaaS resources step, one key to achieve the entire service, this will improve the overall productivity, but also to the complexity of the system, the application of the complexity greatly reduced. In the 2.1 release we plan to support the PAAs platform, I can grow up, the WEB server may load more heavy, we can automatically monitor the load, when you need we can add the machine on your layer. This is my content, thank you.
(Responsible editor: The good of the Legacy)
