云操作系统Nimbula技术白皮书

TECHNICAL WHITE PAPER

Nimbula Cloud Operating System

The Nimbula Cloud OS technology has been designed to uniquely re-spond to the following key require-ments of an enterprise cloud solu-tion:? Scalability – The Nimbula Cloud OS is designed for linear scaling

from a small cluster up to hun-dreds of thousands of computers. This allows an organization to grow and grow quickly.? Ease of use – A highly automated,

hands-off install requiring minimal configuration or interaction dra-matically reduces the complexity of deploying an on-premise cloud. Racks come online automatically in under 15 minutes. Management of cloud services is largely automated, significantly improving operational efficiency.

? Flexibility – The Nimbula technol-ogy supports controlled federa-tion to external private and public

clouds like Amazon Elastic Com-pute Cloud (EC2) as needed by

the customer: during peak times or

for specific applications.? Reliability – With no single points of failure, the Nimbula Cloud OS employs sophisticated fail over mechanisms to ensure system in-tegrity and resilience.? Security – A robust and flexible policy based Authorization System

supporting multitenancy provides mature and reliable security and sophisticated cloud management control.The Nimbula Cloud Operating System is an automated cloud manage-ment system delivering Amazon EC2-like services as either a private cloud behind the firewall or as the basis of a public cloud service.Nimbula’s technology allows customers to easily repurpose their ex-isting infrastructure and build a computing cloud in the trusted envi-ronment of their own data center. Using simple and rapid deployment technologies, the Nimbula Cloud OS transforms underutilized private data centers into muscular, easily configurable compute capacity, quick-ly and cost effectively.

With access to both on- and off-premise cloud services available via a common API, the Nimbula Cloud OS combines the benefits of capital-izing on internal resource capacity and controlled access to additional external compute capacity.

An all star team

Nimbula is a software compa-ny founded by Chris Pinkham and Willem van Biljon – the former Amazon executives that initiated and led the de-velopment of Amazon’s EC2 Public Cloud service.

Developed by a world class team of software developers, including senior former EC2 engineers, the Nimbula Cloud OS provides a uniquely sophis-ticated and nuanced solution for managing the cloud needs of Enterprise customers.Nimbula is backed by a solid team of leading investors, in-cluding Accel Partners and Se-quoia Capital.

For more information on Nimbula? Director

Visit our website at https://www.360docs.net/doc/d41339528.html, or contact sales@https://www.360docs.net/doc/d41339528.html,.

Key Features

The Nimbula Cloud OS provides the following set of key features:

?Self-healing and self-organizing control plane– Robust fail over mechanisms, including monitoring of services and nodes and auto-mated service replacement, ensure system reliability and resilience.?Flexible and powerful group-based Authorization Service– Fine-grained permissions manage-ment based on policy supports advanced access control of multiple users, groups, and customers. The delegation controls allow permis-sioning activity to be executed by the people responsible for the ob-jects and their usage by end users. Collaboration within and across cus-tomers becomes simple to imple-ment, secure, and controlled.?Multi-tenancy– The Authorization Service allows multiple customers, groups and users to co-exist in iso-lation from each other or share re-sources on a single site.?Federation– The Federation Serv-ice supports request forwarding to external sites, such as public clouds, subject to Nimbula’s fine-grained permissions management. This au-thorization filter facilitates powerful control and access management via a uniform API interface to both local and remote private and public clouds.

?Sophisticated placement– In-stance placement can be specified with respect to the type of machine on which the instance should run and the proximity to other instances,

including whether instances should

run on the same or a different node

or cluster. This is useful when pro-

viding redundancy, where for ex-

ample it wouldn’t be desirable to

launch a fail over database server

instance on the same physical ma-

chine as the original database serv-

er. Instance proximity also impacts

on network performance with close

instances providing lower-latency,

higher bandwidth and more reliable

connectivity.

?Persistent storage service– The

Nimbula disk Service allows admin-

istrators to define multiple tiers of

storage for end users to leverage via

self-service volume creation activi-

ties. With a huge compatibility ma-

trix due to the support of NFS back

ends, many combinations of price/

performance/reliability can be of-

fered allowing for the delivery of tar-

geted service offerings for a wide va-

riety of applications and customers.

?Integration with existing user serv-

ices– Nimbula’s Authentication

Service supports Active Directory,

facilitating hassle free user manage-

ment and the efficient reuse of exist-

ing corporate user databases.

?Advanced networking– Besides

supporting flat standard IP alloca-

tion, Nimbula Director allows cus-

tomers to create and declare their

own virtual ethernets. This enables

the launching of instances in multi-

ple isolated layer 2 networks where

customers may provide their own

DHCP server and other layer 2 serv-

ices, such as multicast broadcast

and non IP ethernet protocols. Fur-

thermore, a distributed NAT service

provides a controlled self-service

mechanism for end users to put

cloud instances on the public inter-

net.

?Network security groups are a scal-

able, self-service, distributed firewall

that enforces network policy be-

tween instances and externally that

is completely independent of net-

work vendor, network topology, and

geographic distance.

?Multi-site management – Nimbula

Director clouds can span multiple

geographic locations with a single

administrative interface and central-

ized accounts and policies.

?Metering – Chargeback to the cloud

users is possible with fine-grained

metering capabilities.

?A clean ‘network’ API – A RESTful

HTTP API with JSON encoding sup-

ports a rich set of functionality and a

simple and robust interface to cloud

resources.

?Integrated system metrics and re-

porting – All requests, incidents and

events are subject to comprehensive

recording and reporting, providing a

rich audit trail and detailed graphical

summary of cloud status in near real

time.

?All KVM compatible operating sys-

tems supported – The KVM hyper-

visor provides support for compute

instances running a wide variety of

operating systems, including Micro-

soft Windows, Linux, Solaris, and

BSD. Existing applications running

on diverse operating systems can

easily be moved to the cloud as

Nimbula supports a range of guest

operating systems.

?RedHat and CentOS flavours of the

cloud base available.

Dramatically improving the scale and efficiency of server and application provisioning while reducing applica-tion deployment cycle times and ac-celerating innovation, the Nimbula Cloud OS provides the following key benefits:

?Reduce application deployment time by allowing virtual compute in-stances to be launched in minutes as opposed to hours or days.?Increase infrastructure utilization

rates especially where load is vari-

able.

?Increase the rate of innovation

by providing new opportunities for

innovation through increased infra-

structural flexibility.

?Reduce operational overhead by

dramatically reducing the demand

on data center system administra-

tors.

?Controlled integration with public

cloud services by allowing custom-

ers to tightly control the deployment

of services into external clouds.

1 A hypervisor allows multiple operating systems

to run concurrently on a host computer effectively

providing hardware virtualization. Key Benefits

How it works

The Nimbula Cloud OS abstracts the underlying technology to provide a coherent view of a completely auto-mated virtual data center. Nimbula’s intelligent cloud control software iso-lates customers from the operational and hardware complexity associated with deploying compute in a static pri-vate data center.

Facilitating the rapid deployment of en-terprise grade compute into the cloud and dynamically managing workload allocation, the Nimbula technology transforms data center management.

A RESTful HTTP API provides a simple and comprehensive interface to all as-pects of cloud resource control. Cloud resources can also be managed via a command line interface (CLI) and web control panel, built on top of the API. Beneath the virtual data center abstraction sits a physical layer of stor-age, network and compute hardware managed by multilayer control soft-ware. Nimbula integrates a hypervisor1 (KVM) with network and node man-agement software on each node to achieve automated deployment and configuration.

Compute Control

The cluster is the compute backbone of the Nimbula Cloud OS and consists of a number of x86 based computers, referred to as nodes, connected to a network. All nodes are controlled by the Infrastructure Controller (IC) which ensures that all services run correctly

across the cluster at all times. It is the

IC, which runs as a distributed serv-

ice across all nodes, that enables the

cluster to be self-healing and self- or-

ganizing. At any one time one node is

designated as the IC master, a number

are designated as IC submasters and

the rest act as IC workers. The IC mas-

ter delegates tasks to IC workers to

start and stop services and along with

IC submasters, receives notifications

of service state changes. In the case

of the master failing, the submasters

quickly become aware of the failure

and elect a new IC master, ensuring

system resilience.

Network Control

? The Nimbula Cloud OS facilitates the

creation of highly dynamic virtual

network topologies, independent of

the underlying network topology.

?Network security groups enable in-

stance isolation by providing a dis-

tributed firewall.

? Users are able to dynamically create

virtual ethernets (vEthernets) using

existing networking and associate

these with instances using virtual

NICs (vNICs). Full layer 2 function-

ality, including broadcast, multicast

and non-IP traffic is supported.

? A distributed NAT service provides a

controlled self-service mechanism

for end users to put cloud instances

on the public internet.

Storage Control

The Nimbula disk Service allows ad-

ministrators to define multiple tiers

of storage to deliver varying levels of

price / performance / reliability. They

delegate the permission to use these

tiers to customers, groups, and end

users to leverage via self-service vol-

ume creation activities. Cloud users

create their own persistent volumes

and decide which of their instances

mount each of those volumes. Shar-

ing and delegation of the volumes are

made possible by Nimbula Director’s

collaborative permissioning service.

Monitoring

The Nimbula Cloud OS gathers and

collates monitoring information that

can be accessed via a web interface

and integrated with monitoring soft-

ware.

EC2-like services without

the risk

Cloud computing offers the promise

of highly scalable infrastructure with

greater flexibility and dramatically

lower costs. The Nimbula Cloud OS

brings these values of scale, low fric-

tion and lowest cost to the private data

center. Differentiated by high levels of

automation, flexibility, control, scalabil-

ity, security and resource accountabil-

ity, the Nimbula OS provides the com-

plete utility-grade cloud management

solution.

Copyright? 2011 Nimbula, Inc. All rights reserved. This product is protected by U.S. and international copyright and intellectual property laws.