IBM Middleware in the Clouds

Seema Kumar Product Manager, WebSphere Application Infrastructure seemakumar@in.ibm.com

Agenda
Business Perspective
Whats driving customer costs? Role of Cloud Computing and Pattern-based middleware

Technical Perspective
Building blocks for cloud Virtualization & Automation Dynamic Infrastructure Services

Conclusions

Building Clouds with WebSphere A Business Perspective

What is Driving Customers Costs?

Labor Costs Number of applications drives labor costs
Typical enterprise customer has hundreds of apps The installation, maintenance and problem determination for the applications entire stack is done application by application

Hardware utilization rate drives hardware,

power and space costs Non-virtualized customers: each app runs on its own server
Source: GTS Admin Time Study

Note: This study did not include database administrators

Virtualization can drive the average utilization rate from 10% to 70%

Other factors
Age hardware and facilities Data center location

How Can Customers Drive Down their Operational Costs?

Drive up utilization rates by moving

Hardware Utilization their distributed workload to a private cloud (hardware as a service)
Application Application Application Application Application

Application

Use image management software to

drive down operating system and middleware upgrade and maintenance costs Standardize on a small set of hardware and middleware configurations Middleware Operating System Hardware Image Management
Application Application Application Application

Upgrade / Maintain

Middleware Middleware Middleware Middleware

Application

Middleware

Centralize logging for all

Problem Determination

Operating
Hardware System Hardware Hardware Hardware

components within the stack Scrape logs for known symptoms and trigger PD tools pre-emptively

Hardware

An effective Cloud Computing deployment is built on a dynamic application infrastructure and is highly optimized to achieve more with less.

CLOUD COMPUTING

VIRTUALIZATION

OPTIMIZATION

STANDARDIZATION

AUTOMATION

Reduced Cost

Increased Flexibility

leveraging virtualization, optimization, standardization and automation to free up operational budget and enable more business investments

A services oriented approach to applications, middleware and infrastructure

Collaboration Business Processes Industry Applications

CRM/ERP/HR

Applications
Process Engine Business Rules Application Server

Analytics Database

Portal

Middleware
Data Center Fabric

Servers

Networking

Storage

shared - virtualized - dynamic

Infrastructure Cloud is a consumption and delivery model for many types of business services, in which the user sees only the service, and has no need to know anything about the technology or implementation.

Building Clouds with WebSphere An Infrastructure Perspective

WebSphere in the Clouds

WebSphere & Cloud Computing Building Blocks for Clouds Hybrid Cloud Connectivity

Middleware Services

Enable customers to: - Accelerate time to value - Reduce cost of owning and operating enterprise applications and middleware - Capture new & evolving business opportunities with improved agility

Characteristics of a Cloud

WebSphere & Cloud Computing

Building Blocks for Cloud

- More Responsive: Dynamically allocates resources to meet demands - More Optimized: better utilizes system resources and lowers TCO - More Agile: better aligns IT capabilities with business needs - More Resilient: prevents, isolates, and recovers from failures

Pre-Cloud middleware was - Scalable: Add additional resources to meet demands - Available: Redundancy to avoid outages - Consolidated: Shared hardware resources

Application Infrastructure Evolution

WebSphere & Cloud Computing

Building Blocks for Cloud

- Silod - Decentralized - Dedicated

Standardize, Consolidate, and Virtualize

- Centralized - Shared

Automate, Enlighten, and Scale

- More Responsive - More Resilient - More Optimized - More Agile Hybrid Integration

Corporate Firewall

Evolving application infrastructure:

1. 2. 3. 4. 5. Virtualization, standardization, and automation Cloud-enabled applications that are scalable and optimized Integrated services across the enterprise and across clouds Intelligence & autonomics for responsive & resilient systems Self-service and management optimizations for better agility Public Cloud - Fast time to value - Low barriers to entry - as-a-Service delivery

Building Clouds with WebSphere

WebSphere & Cloud Computing

Building Blocks for Cloud

eXtreme Scale, Compute Grid

Scalable & Optimized OLTP & Batch applications

Cloud-enabled Applications

Dynamic Infrastructures

Elastic & resilient systems aligned with business objectives

Virtual Enterprise, eXtreme Scale, DataPower

Application Server

Application & Runtime governance

Runtime Management

Security Services

Identity, access, & transport security management

Application Server, Datapower

Cloudburst Appliance

Cloud shaping, Virtualization + Topology Automation Management

Connectivity Services

Integrating services across enterprises

Cast Iron, ESB, Datapower, MQ

WebSphere delivers proven technologies & services to build clouds

- Challenge: Assess the maturity of each pillar, determine how to incrementally pursue the cloud. - Its more than just buying products, customers must establish operational & app dev discipline

Enhanced with the broader IBM software portfolio

WebSphere & Cloud Computing

Building Blocks for Cloud

Application Developer, Software Architect

WebSphere, eXtreme Scale, Compute Grid Cloud-enabled

DB2 Pure Scale

Applications

Dynamic Infrastructures

Virtual Enterprise, eXtreme Scale, DataPower

Provisioning Manager

Monitoring, Netcool Omnibus, Netcool Impact

Application Server

Systems Management

Security Services

Application Server, Datapower

Identity Manager, Access Manager, ISS Proventia

Rational Automation Framework for Cloudburst WebSphere Appliance Provisioning Manager, Service Automation Manager

Virtualization + Automation

Connectivity Services

Cast Iron, ESB, DataPower, MQ

WebSphere combined with the broader IBM software portfolio delivers enhanced cloud capabilities

Intelligent Management in the Cloud

Tivoli Monitoring (ITM), Service Automation Manager (TSAM) Provision & Monitor WebSphere Cloudburst Appliance (WCA)

WebSphere & Cloud Computing

Building Blocks for Cloud

Cloud Shaping & Topology Management

Dynamic WebSphere Infrastructure Virtual Enterprise Services (WVE)

WebSphere Application Server

Application

DB

Optimized with WebSphere eXtreme Scale (WXS)

- More Responsive: WVE, WCA, and TSAM can dynamically allocate resources to meet demands - More Optimized: WXS & WVE combined with ITM better utilizes system resources and lowers TCO - More Agile: WVE, WXS, & WCA better aligns IT capabilities with business needs - More Resilient: WVE & WXS prevents, isolates, and recovers from failures

Middleware Services in the Cloud

- Offer ready-to-run WebSphere products as cloud services - Lowers barriers of entry and allows customers to get started quickly - Ideal for prototypes, proof-of-concepts, development, and test

WebSphere & Cloud Computing

Middleware Services

Virtualization & Automation

eXtreme Scale, Compute Grid

Scalable & Optimized OLTP & Batch applications

Cloud-enabled Applications

Dynamic Infrastructures

Elastic & resilient systems aligned with business objectives

Virtual Enterprise, eXtreme Scale, DataPower

Application Server

Application & Runtime governance

Runtime Management

Security Services

Identity, access, & transport security management

Application Server, Datapower

Cloudburst Appliance

Cloud shaping, Virtualization + Topology Automation Management

Connectivity Services

Integrating services across enterprises

Cast Iron, ESB, DataPower, MQ

Customer Pains

Customer Pains
Takes too long to create middleware infrastructures Manual and error-prone process

Ways to address the pains

Building homegrown WebSphere Admin Scripts WebSphere Cloudburst Appliance Integrated solution with WebSphere Cloudburst Appliance, Rational Automation Framework, and Tivoli Management Products

What admin & management efficiencies can be made?

The average lead time to get a new application environment up and running is 4-6 weeks
Approvals, procurement, shipment, HW installation, license procurement, OS installation, application installation, configuration

30% of bugs are introduced by inconsistent configurations

These bugs are often of the most difficult variety to detect They often emerge when moving between dev/test, QA, production

Because its so expensive to set up an environment, there is an incentive to hold onto them even when no longer needed just in case.
Future environments = new hardware, instead of recycling returned hardware, and this takes time and money

Application Configuration & Provisioning Characteristics

1. Speed - build out environments quickly
A new WebSphere cell should take you minutes, not weeks

2. Repeatability - automation to avoid redundancy

Your 2nd, 3rd, and Nth environment should be faster to build than your 1st environment

3. Consistency - limit the variables across deployments

Your configurations should get seamlessly promoted from test -> pre-production -> production
Application

WAS Web Server Description of the Middleware Topology

Resource Definitions

Automation Magic

DB
Web Server WAS
Resource Definitions Application

Test Env Variables

Production Env Variables

2009 IBM Corporation

Pattern-Based Deployment of Middleware Services in the Cloud

Selects a pattern based on application requirements

WebSphere Cloudburst Appliance (WCA)

Creates & customizes patterns of middleware services

Application Developers

Instantiates & manages patterns of middleware services in the cloud

Admin

Access applications in the cloud

middleware services on virtualized infrastructure

WAS

DB2

Application Clients

Customers manage patterns of middleware services, and no longer need to deal with the details of middleware installation & configuration leading to quicker time-to-value, improved consumability, and lower costs

WAS product team accelerates Agile development practices with Websphere CloudBurst Appliance Client Pains
Needed to enable rapid access to WAS topologies to improve quality and decrease costs Inefficient use of infrastructure resulting in low hardware utilization

Real Results
$500K in direct savings, $2.1M in enabled efficiency gains through Agile development practices in the first year of deployment Reduced topology install time from 3 Hours to 20 Minutes Increased hardware utilization from approx 10% to over 60%

Dynamic Infrastructure Services

eXtreme Scale, Compute Grid

Scalable & Optimized OLTP & Batch applications

Cloud-enabled Applications

Dynamic Infrastructures

Elastic & resilient systems aligned with business objectives

Virtual Enterprise, eXtreme Scale, DataPower

Application Server

Application & Runtime governance

Runtime Management

Security Services

Identity, access, & transport security management

Application Server, Datapower

Cloudburst Appliance

Cloud shaping, Virtualization + Topology Automation Management

Connectivity Services

Integrating services across enterprises

Cast Iron, ESB, DataPower, MQ

Customer Pain
1. Static Middleware Infrastructure
Doesnt react well to spikes in demand Resources are under-utilized Not well-aligned with the business

The infrastructure should manage provisioning application and middleware resources to achieve some stated business level objectives.

2. Fragile Middleware Infrastructure

System cant detect that a failure will probably occur Failures arent isolated, and impact more than it should

The infrastructure should monitor and react to conditions that effect the health of the cluster-member JVMs

Traditional Middleware Architecture

-Application clusters are statically defined -Low CPU utilization -Many JVMs to Manage -Very little insight to cluster-member health
firewall Web DMZ firewall Trusted Zone

Application 1 Cluster

Application 2 Cluster

HTTP Traffic
Web Server Tier

Application 3 Cluster

= JVM
Middleware Server Tier

Policy-based Management
-ODR is responsible for: -starting/stopping JVMs to meet demand -Ensuring higher priority applications/users/etc are serviced first (via flow control) -Administrators describe requirements (min/max application instances, etc) through policies -Resource-driven load-balancing within clusters -Keeping track of application/user/etc resource usage (for chargeback)
Business-level Service Agreements

HTTP Traffic
Web Server Tier On-Demand Router Tier (ODR)

Pool of Resources

Applications

Optimizing the Pool of Resources with Dynamic Clusters

-Applications are installed to a Dynamic Cluster -ODR uses policies to determine when to start/stop dynamic cluster members -Dynamic Cluster member = WAS JVM
App 1 Dynamic Cluster Min = 1, Max=2 Frame 1 Frame 2

HTTP Traffic
Web Server Tier On-Demand Router (ODR)
App 2 Dynamic Cluster Min = 1, Max=2

= WAS JVM

WVE- Some examples

- A memory leak in the JVM could be detected and reacted upon - A spike in demand could trigger additional JVMs be started - Decisions can made, and actions taken without Admin intervention
Web Server
Memory Leak Event

Health Mgmt

Ops Monitoring

App Editions

Memory Leak Detected!!

WAS WAS Dynamic Cluster JVM JVM

ODR

React to prevent a problem!

WAS WAS Dynamic Cluster JVM JVM

Spike in demand!

Health Management Health Policies

Helps mitigate common health problems before production outages occur
Health policies can be defined for common server health conditions

Health conditions are monitored and corrective actions taken automatically

Notify administrator Capture diagnostics Restart server

Application server restarts are done in a way that prevent outages and service policy violations

Health Conditions Age-based: amount of time server has been running Excessive requests: % of timed out requests Excessive response time: average response time Excessive memory: % of maximum JVM heap size Memory leak: JVM heap size after garbage collection Storm drain: significant drop in response time Workload: total number of requests

Cloud-Enabled Applications

eXtreme Scale, Compute Grid

Scalable & Optimized OLTP & Batch applications

Cloud-enabled Applications

Dynamic Infrastructures

Elastic & resilient systems aligned with business objectives

Virtual Enterprise, eXtreme Scale, DataPower

Application Server

Application & Runtime governance

Runtime Management

Security Services

Identity, access, & transport security management

Application Server, Datapower

Cloudburst Appliance

Cloud shaping, Virtualization + Topology Automation Management

Connectivity Services

Integrating services across enterprises

Cast Iron, ESB, DataPower, MQ

Application Challenges
The cloud provides a dynamic & scalable infrastructures on highly virtualized hardware Applications must:
Scale well to take full advantage of the cloud Be machine independent, so they can run on the best available machine in the cloud Container-managed, so they can be optimize by the infrastructure Have a small memory footprint

Applications probably:
Dont scale well Write to local log files, read from local configuration files, etc Are not optimized for container-managed services
Home-grown batch Home-grown or inefficient caching

Cache lots of data locally, resulting in duplicate cached data cross many JVMs, which leads to memory-bound systems, instead of CPU-bound systems

Linear Scaling
- Applications should scale linearly - Bottlenecks in data access, logging, and application state management prevent applications from scaling - Dynamic Infrastructure Services provide AutoScaling features for applications -if apps dont scale, they cant take advantage

Throughput vs. Load / Resources

Saturation Point
Linear Scaling Non-linear Scaling Bottleneck Load / Resources

Reducing the Average JVM Footprint with a Distributed Cache

With virtualization & consolidation, customers arent running out of CPU, they are running out of memory! Reduce the JVM memory footprint with smarter caching

Large app JVMs with duplicate cached data

App Cache

App Cache

App Cache

App Cache

Smaller app JVMs that connect to a faulttolerant, centralized distributed cache

App

App

App

App

WebSphere eXtreme Scale cache

Cloud-wide (or Enterprise Wide) Data Cache

Expose cached data to all applications running within the cloud Improve fault-tolerance, performance, and reduce the Application memory footprint App 3 App 2 App 4

App 1

WebSphere eXtreme Scale Cache

App 5

Web Application Challenges

1. Web Application Infrastructures dont scale well 2. New application infrastructure designs enable scaling to tolerate huge transaction volumes
1. Think: Facebook, Google, Amazon, Ebay

3. WebSphere eXtreme Scale delivers technology to build highly scalable web application infrastructures

Traditional Application Infrastructures

Add stuff to MY shopping cart

Route Request to JVM that holds MY shopping cart Add stuff to a shopping cart My Shopping Cart Contents

Web Server

Application Conversational State App Server

Application Conversational State App Server

- Business Transactions and Conversational State are located in the same JVM - Conversational State (HTTP Sessions) require session affinity & special failover - App Server required complex HA and failover - As a result, scalability was limited

New Application Infrastructures

Add stuff to MY shopping cart

Web Server
Add stuff to a shopping cart

WVE
Application App Server Auto-scale

Application App Server Conversational State

WebSphere eXtreme Scale

My Shopping Cart Contents

Conversational State
WebSphere eXtreme Scale

- Business Transactions and Conversational State are in separate tiers - Business Transactions are stateless - Conversational State is fault tolerant - Each tier can be scaled independently - No scalability limitations

Connectivity Services

eXtreme Scale, Compute Grid

Scalable & Optimized OLTP & Batch applications

Cloud-enabled Applications

Dynamic Infrastructures

Elastic & resilient systems aligned with business objectives

Virtual Enterprise, eXtreme Scale, DataPower

Application Server

Application & Runtime governance

Runtime Management

Security Services

Identity, access, & transport security management

Application Server, Datapower

Cloudburst Appliance

Cloud shaping, Virtualization + Topology Automation Management

Connectivity Services

Integrating services across enterprises

Cast Iron, ESB, DataPower, MQ

Why Integrate?
CRM
Marketing
I need to do a credit check

Shipping
Where are my orders?

Sales

? ? ?
What are my hottest leads? ?

Has this customer paid?

Where are my invoices?

Finance

Why Integrate?
CRM
Marketing Shipping

Sales

Hottest Leads

Orders

Invoices

Finance

39

Cast Iron Systems

Thousands of Customer Integrations Connecting Cloud & Enterprise
Enterprise

Applications

Platforms

Infrastructure

Cast Iron Systems

Thousands of Customer Integrations Connecting Cloud & Enterprise
Enterprise

Applications

Platforms

Infrastructure

End-users :: Productivity End-users Productivity Cloud Providers: Loyalty Cloud Providers: Loyalty

Value Value

End-to-end connectivity

Complete Flexibility

Total Connectivity

Complete Reusability

For All Cloud Integrations

TIP Exchange Cast Iron Cloud2

Mashups

TIP Development Kit Physical Appliances

Synchronization

Virtual Appliances

TIP Community

Migration

42

Conclusions

An effective Cloud Computing deployment is built on a dynamic application infrastructure and is highly optimized to achieve more with less.

CLOUD COMPUTING

VIRTUALIZATION

OPTIMIZATION

STANDARDIZATION

AUTOMATION

Reduced Cost

Increased Flexibility

leveraging virtualization, optimization, standardization and automation to free up operational budget and enable more business investments

Q&A

Cloud Ecosystem

Independent Software Vendors (ISVs)

- Enable ISVs to: - Build applications for the cloud - Exploit new channels to deliver their applications - Leverage as-a-service delivery and business models
Collaboration/Content/Portal
develops ISV

CRM Enterprise Resource Management Business Intelligence & Analytics Etc

Application
Delivered via IBM channels

WAS

WAS

DB2

DB2

Customer Datacenter

Public Cloud

Systems Integrators (SIs)

- Enable SIs to: - Sell billable services to help customers create private clouds - Build hosting capabilities and provide as-a-Service offerings Private Cloud
Enablement of capabilities for private clouds

Dynamic Infrastructures Cloud-enabled Apps Connectivity Security Centralized Management Virtualization + Automation
Time WebSphere eXtreme Scale Cast Iron WebSphere Application Server

WebSphere Virtual Enterprise WebSphere Compute Grid WebSphere MQ WebSphere DataPower

WebSphere Cloudburst Appliance