Data center development prototype --ENIAC
ENIAC is short for electronic numerical integrator and computer Electronic Numerical Integrator and Computer, designed for the US Ballistic Research Laboratory storage firepower table was developed in 1946, to participate in development work is based on Moore School of Electrical Engineering, University of Pennsylvania the Moxi Li and Eckert led the research team.
ENIAC also has many disadvantages: In addition to large volume, multi-power outside, due to the high heat generated by the operation of the machine that the tubes are easily damaged. As long as there is a tube is damaged, the entire machine can not operate normally.
Data center virtualization technology commercialization --TRADIC
first transistor computer TRADIC Bell Labs successfully developed in 1954, just as the commercial system in time to appear around 1960, then led the mainframe to achieve a breakthrough. Mainly for government and military and development, the environment has unique requirements for space and security systems, as performance requirements, it is desirable to provide additional support performance and resource sharing on the same system, so virtualization technology will gradually enter horizon. The first commercialized application of virtualization technology has been implemented on the 1972 IBM VM / 370 OS.
C / S structure model of computing and the Internet
in the mid-1990s, the Internet had a tremendous impact on the market, providing more options for deploying ten years down the data center, network providers and hosting providers in the hundreds created to develop thousand in the data center as a service model is accepted by most companies.
C / S structure
1. By reasonable allocation of tasks to the Client-Server and an end, reduces the communication overhead of the system, needs to be installed before the client management operations.
2. The server-side main provider of data management, data sharing, data and system maintenance and concurrency control, client program mainly to complete specific business users.
3. Develop relatively easy, easy to operate, but the maintenance and upgrade client applications more difficult.
B / S structure
1.Browser / Server (Browser / Server) structure, along with the rise of Internet technology, a variation of C / S structure or improved structure. ; In this structure, the user interface is fully implemented by WWW browser
2. The client basically no specialized applications, the basic application on the server, upgrade and maintenance of such an application can be done on the server side, and easy maintenance.
Energy problems arise
arise PC boom and the data center, take up more space, increase energy consumption, in 2002, the data center has consumed 1.5% of US energy, and a rate of 10% per year, to this end, data center owners began to deploy more cost-effective, green infrastructure, in 2007, large-scale data center operators began to use recyclable energy technologies, such as wind, solar, etc., to support the daily operations of the data center.
Emergence of modular data center
device data centers are deployed inside the container, therefore, also known as container data center. The most famous include Sun Blackbox-- the container data center 280 servers have been deployed in the container inside the cabinet 20 inches, and can be shipped all over the world. This data center although not grand and imposing traditional data center, but it requires only 1% of the construction costs of traditional data center, with a flexible mobility, but also can significantly reduce deployment cycle.
The modular data center is a new generation of cloud-based data center deployment form, its modular design concept, coupled reduce the maximum infrastructure of the room environment. Integrated power supply and distribution, cooling, rack, air containment, cabling, ring monitoring subsystem, to improve the overall operational efficiency of the data center and quickly deployed, and resilient expansion of green energy.
From the configuration form, it can be divided into micro-modules MDC and container data center products CDC.
Cloud data centers
SaaS demand computing resources to achieve the transformation of the infrastructure needed to bring to the ordering patterns. The business model by means of the network infrastructure and data center operators together to provide the massive growth of data bandwidth resources to users.
The basic building blocks of the data center
What is a data center?
Wikipedia definition: data center is a complex set of facilities. It includes not only computer systems and other ancillary equipment (e.g., communication and storage systems), further comprising a redundant data communications connections, environmental control equipment, monitoring equipment and safety devices.
Google definition: data center is multi-functional building that can accommodate multiple servers and communications equipment. These devices are placed together since they have the same needs and environmental requirements on physical security, ease of maintenance and so disposed ", and" not only a collection of servers. "
Huawei definitions: Enterprise Data Center is RESERVOIR the data center solution is Dongting Lake.
Reservoir Lake + = data Center (calculated data / storage / exchange center)
Pacific + Yangtze + Yellow = pipeline network operators
tributary Chengshiguanwang + = enterprise network channel
taps = network Termination
water = (voice / data / video) streams
modular data centers constitute
the data center is a service-oriented architecture foundation, is to support the company's business operations and future growth of the core safeguards, including the following aspects: a secure network architecture; reliable support facilities (room, generator, UPS, air conditioning, etc.); integrated server / application platform; centralized storage and backup; unified system management platform; operation and maintenance management organization and process-oriented customer service.
the role of the data center: connection the resource server, storage, and data centers across the
computing network: connections User computing resources and
storage networks: Attached Storage
data center interconnect: connection across data center resources
L0: Gailou, referring to civil engineering, computer room building part
L1: Infrastructure: room facilities, refers to room decoration, power, cooling, fire protection, etc.
L2: infrastructure, computer room infrastructure layer, including servers, storage, networking and some virtualization software station
L3: application platform, Web Hosting PASS
L4: business, services
of data center infrastructure development
in 1994, Internet enter China, information technology began to appear, the engine room is no longer just a computer system, information system began in the true sense. .
The twenty-first century, IT systems and information technology the driving force for enterprise development, infrastructure data center IT as a support system, it has become a must.
Now, no longer confined to the data center site infrastructure, its connotation, extension more widely, IT, network, relocation, operation and maintenance are gradually incorporated into the data center construction category
of data center infrastructure trends
of business development enterprise, data center and development development of new technologies / new materials, between the three are complementary, mutually reinforcing each other and limit; at the same time, development of their business is the driving force behind the development of the data center, which determines the development of demand.
In recent years, changes in the data center: from single to multi-direction service,
data center Fusion Module Solution
Product Description:
Huawei fusion module 2000 module is a new generation of intelligent data center infrastructure solutions.
Highly integrated design, the integrated cabinet, all subsystems for power distribution, cooling, and management of the wiring support flexible deployment single or double cold closed / hot channels, modular design, supply and distribution, temperature control, channel of the cabinet, wiring, integrated in a monitoring module, to achieve rapid delivery, deployment requirements demand. To enhance the power supply, temperature control system reliability, energy efficiency, and by locating the convergence alarm, fault isolation, automatic asset management, significantly improve the efficiency of operation and maintenance.
Scenario:
Single module applied to the total power consumption is less than 128kW, an area of 500 square meters below sized data center can also be used in large enterprise branch data centers.
A plurality of modules constructed in large data centers. Can be deployed to meet the needs of government, education, health care, finance, telecommunications and other industries large data centers.
Features:
with integrated integrated, safe, reliable, energy saving and saving room area, the installation saving time, effort, worry, architecture compatible, fast and flexible deployment, improved monitoring, refrigeration stability characteristics, is a new generation of intelligent micro module data center products.
Future data center
contents 1. Data centers should at least include space environment, IT equipment and monitoring and management.
Site environment, to ensure that the information system reliable and stable operation;
IT equipment is an information system on which to run the IT infrastructure;
monitoring and management, including effective management of the site infrastructure and IT infrastructure.
2. "Modular Data Center", is to meet unpredictable business and IT growth in demand; expand the system without interrupting service;
3. "Green Data Center", the data center, the focus is on energy saving, that is, using a variety of techniques and tools, effectively reduce operating costs (emphasis reduce PUE value)
4. "smart data center" is to realize the IT equipment, unified monitoring and management of venues and facilities and IT processes; to support energy management and asset management; support real-time information, simulation and remote monitoring, and other technologies;
5. "cloud data center" cloud computing has its basic characteristics
based on the powerful computing power and storage capacity to support large-scale infrastructure; and
use a variety of virtualization technology to enhance resource utilization rate;
relying on-demand access to elastic scalability support, pay-on-demand and strong versatility;
professional operation and maintenance support and a high degree of automation technology
key evaluation indicators green data Center
data Center energy efficiency indicator (PUE)
PUE = total data center use electricity consumption / IT equipment energy consumption
ideal PUE = 1.6-2.0 even less common PUE = 2.0-2.5 To a higher
data center based composition provided
a data center infrastructure is a field comprising a plurality of integrated subsystems, FIG Huawei under the definition of a data center infrastructure layer L1 multiple systems.
Typical traditional data center logical partition
cloud data center infrastructure Concepts
DC: Data Center, data center. DC is a physical concept, refers to the information to achieve centralized processing, storage, transmission within a physical space (eg room), switching and management. Server, storage and network devices are DC key equipment, power, cooling, fire protection, monitoring and other infrastructure is key to supporting DC.
POD: Point of Delivery, distribution point. To facilitate resource pool DC operation, a DC can be divided into one or more physical partitions, each physical partition called a POD. Thus, POD is the physical concept, a DC basic deployment unit, a physical device can belong to a POD.
AZ: Availability Zone, the available area. AZ is a logical concept, represents a fault isolation region. For example, some hosts share a set of power and network facilities, when this facility fails, this part of the resources are not available on all. When planning, AZ and DC according to the actual deployment of flexible mapping. For example: In a large-scale public cloud, AZ may include a plurality of DC; in small and medium-scale private cloud, a DC can be set within several sets of independent AZ; of course, can also be a DC plan is a AZ, then DC and AZ are equivalent.
Cloud data center VDC and Tenant
VDC: Virtual Data Center, the virtual data center, a tissue collection VDC is available resources, including general computing, storage and network resources.
Tenant: Tenant, created by the system administrator and distribution. VDC is a tenant of the actual owners and managers, different VDC corresponding to different tenants.
➢VDC similar in form to a physical data center, is the physical data center performance virtualization layer.
➢ In the public cloud scenario, the system administrator can define and assign tenants VDC VDC, VDC is the only tenant management resources available under the VDC.
➢ In the private cloud scenario, VDC can be defined flexibly assigned to one business / applications / departments. The system administrator can manage different levels of resource quotas for different services / applications / departments within the enterprise by VDC.
VPC cloud data center
VPC: Virtual Private Cloud, Virtual Private Cloud.
VPC using VDC resources, a VPC can only belong to a VDC, and may comprise a plurality of VDC VPC. Each VPC is a secure domain, corresponding to a service / application / department. VPC offers the following features:
➢ isolation environment: VPC provides isolation of virtual machines and network environment, network isolation to meet the requirements of different services / departments.
➢ rich business: Each VPC provides independent vFW, vLB, security groups, EIP, IPsec VPN, NAT and other services.
➢ flexible networking: VPC available networks, and internal network routing and other networking modes Direct
vRouter
Layer interworking gateway services as vRouter subnet for subnets.
Subnet
Subnet used to isolate Layer broadcast domain, corresponding to a subnet segment.
VPC different Subnet within the same three-layer gateways, all on the same vRouter.
Subnet same default interworking; Subnet default interworking between different, can also be isolated by configuring the security group.
vFW
vFW the VPC as the boundary, in addition to providing secure access control within VPC external access, access may also be provided within the external access VPC.
Properties available are: FW, EIP, SNAT, IPsec VPN and so on.
VLB
VLB for load balancing capabilities between the foreign provide the internal server.
VLB can take a plurality of servers, the user can apply them to multiple different servers.
DCN Fabric cloud data center network
fabric: a set of leaf nodes interconnected in a spine composed physical network topology data base center
spine: a backbone node, VXLAN fabric core network node, to provide high-speed IP forwarding, each connected via a high speed interface function leaf nodes.
leaf: leaf node, the access node VXLAN Fabric network functions, providing access to various network devices VXLAN network functions.
Fabric physical network
CE series physical / virtual switch device as Fabric carrier
no difference between access nodes
flat structure
flexible networking
easily scalable
Fabric virtualization
Fabric VXLAN using virtualization technology
supports pure hard / mixing the Overlay
Fabric Management
SDN Control Overlay, automatic deployment of virtual network
Underlay network is L2 / L3 network, IGP routing protocol
BGP-EVPN Overlay as a control plane of
the service provided by the Fabric
provided between the access node L2, L3 visits
Cloud data center trends in the evolution of
enterprise IT cloud transformation trend
with the continuous progress of information technology, data centers are becoming more and more rigid, isolated, management more complex, more and more inefficient business ...... transformation is imminent.
IT enterprise cloud technology, enterprise IT transformation of key
IT cloud based, organic combination of the Internet, the traditional IT business gradually, step by step evolution to the cloud-driven business model. Private cloud is the key to the current restructuring of
the industry of private cloud development status and trends
1. After laas based services, network services more and more mature
2. majority support OpenStack
3. Self-service increasingly demanding
4. more convenient access public cloud, private cloud and public cloud provides integrated management
5. vendors to support more levels of
cloud process is divided into three stages, virtualization-based service-oriented evolution 1.0 times 2.0 times to 3.0 times It is based on the business architecture of cloud cloud Native architecture of the era. The current development of the industry Private Cloud 2.0 has entered the stage, IaaS +, OpenStack, service gradually for developers, hybrid cloud management.
Most companies are in the evolution of virtualization to private cloud process
NIST definition of private cloud has five characteristics: on-demand self-service, broad network access, resource sharing, rapid scalability or extended service and measurable.
This transformation process is gray, it also gives the process of transformation of our business adds a lot of uncertainty and confusion.
Enterprise IT transformation stage in confusion
: virtual chimney, resources can not quickly adapt business management system is not uniform, internal processes can not match the demand;
the deployment of private cloud generally remain in the traditional data center management software + virtual stage.
The driving force behind the transformation of enterprise
cost reduction, flexible business online, increase reliability. We summarized as: efficient, quick, and reliable.
Huawei's cloud data center
for the future of cloud data center architecture fusioncloud
provide solutions ranging from infrastructure to the cloud to cloud management platform to meet the final transformation of enterprise cloud cloud services in all aspects of the various business scenarios involved.
Under OpenStack architecture building convergence of computing resources pool
resources integration is the first step in resolving the current virtualization chimney.
To build a framework developed by OpenStack architecture, we first solve the problem of unified computing resource management. Supports physical machine, Vmware, FusionSphere and KVM (NFVI scene) unified monitoring, distribution and recycling.
Based storage services and integration of heterogeneous application template
followed by the integration of storage resources. Heterogeneous and different types of storage fusion, while completing service requirements for different storage performance, to achieve the storage of SLA
SDN support services across data center automation and flexible deployment
of network resources must also be opened automatically configure and implement cross-domain service deployment through automation and network big story. The solution covers the entire physical network, virtual networks and wide area networks.
Unified service access platform
unified cloud services access platform, fast access and flexible cloud orchestration services. Distributed cloud data center management, service-minute stage to complete automation across the data center deployments.
Based on flexible business VDC
VDC across resource pools to compartmentalize resources, VDC match the customer's organization is decentralized Fenwick cloud resources. VDC has a self-operational capacity, allocating cloud resources of independent business units, focused business, cloud platform maintenance is the responsibility of the IT department, use and construction of the separation improve the ability of business-driven resource.
SLA scheduling strategy to meet the business needs
of different business indicators for the resource requirements of performance and function are different, the premise of the perception of business, by the division of resources SLA levels can be achieved for different strategies of business required for different services different allocation of resources.
Based on business-driven elastically stretchable and service scheduling
in the course of running the business, for the resource performance, number of appeals is dynamic, FusionCloud architecture after the perceived demands of the business, may initiate appropriate strategies to achieve resilient and elastic resources or scheduling .
Unified management of hybrid cloud
unified: unified lightweight hybrid cloud management service projects
open: Support for OpenStack cascading support AWS S3, EC2
flexible: traffic load automatically across the cloud stretching across cloud virtual machine migration
cloud disaster recovery solutions
➢ no double gateway live, live traffic performance bis 30%
➢ visualization disaster, a disaster recovery button, business recovery time of from 30 minutes to 10 minutes
➢ deduplication backup public cloud backup
disaster recovery program series, is a traditional IT & cloud data in under different scenarios centers, protection programs for business customers offered.
7 disaster recovery sub-programs under the traditional IT environment to meet customer demand for different levels of disaster recovery. Among HyperMetro characteristics play the main active-active data center solutions to meet customer business interruption 0, 0 data loss disaster highest level requirements, and support cloud of a smooth upgrade from the traditional to the IT environment.
Private cloud disaster recovery program is based on the architecture of the disaster recovery business openstack ability to service orchestration, currently provides backup services, standby service (the industry's first cross-architecture support openstack center disaster recovery); as a double live the highest level of service.
Questions:
data center modular constitute what?
Solution: secure network architecture; reliable support facility (engine room, generator, UPS, air conditioning, etc.); integrated server / application platform; centralized storage and backup; unified system management platform; operation and maintenance management organization for customer service and processes.
What is the relationship between the cloud and the data center VDC VPC that?
Solution: VDC: Virtual Data Center, the virtual data center. VDC is the set of available resources of an organization, including general computing, storage and network resources.
VPC: Virtual Private Cloud, Virtual Private Cloud. VPC using VDC resources, a VPC can only belong to a VDC, and may comprise a plurality of VDC VPC. Each VPC is a secure domain, corresponding to a service / application / department.
