صفحه 1:
۲ 5۱۱۵۲۱۲۵
حه» . 10-5 ظا 813,۵07
shlomo.swidler@gmail.co
صفحه 2:
Developing Cloud
Computing Applications with
Java
Overview of Cloud Computing
Amazon’s Cloud Platform
Google’s Cloud Platform
Application Development Challenges Posed by
the Cloud...
... and Java Solutions
1۱ ees oe
صفحه 3:
ecm ME ava Technology Day 5:
About Me
| تا رت رها
Music marketing on social networks
Patent-pending targeting technology
Java, MySQL, auto-scaling & cloud-based
Active in the Cloud Computing community
Open Cloud Computing Interface Working Group
(an Open Grid Forum initiative) participant
Contributor to Open Source cloud & Java projects
Se eee eects
صفحه 4:
Cloud Computing Is...
A style of computing in which dynamically
scalable and often virtualized resources are
provided as a service over the Internet. Users
need not have knowledge of, expertise in, or
control over the technology infrastructure in
the “cloud” that supports them.
- Wikipedia
eee ee کح 0
صفحه 5:
Cloud Computing Is...
ار ۱۱۱۲۰ لك فت انا
scalable and often virtualized + ces are
provided 3 54 over the Internet. Users
need not have knowledge of, expertise in, or
control over the technology infrastructure in
the “cloud” that supports them.
- Wikipedia
لا ١ طاناا يها دمل لعب Deere et
صفحه 6:
ااا الس مد
Cloud Computing ls...
Cerone teeter
صفحه 7:
ااا الس مد
Cloud Computing Is...
« Infrastructure as a Service — laaS
« Platform as a Service — Paas
« Software as a Service - Saas
eee en et et
صفحه 8:
ام ا الس مد
Cloud Computing Is...
Computing Resources
Infrastructure Platform تست
Processor يتا >اع 568 هالاها
Memory (۷/۷۳ CRM System
Storage Python VM SNe)
Network MapReduce SCM System
ودره ۳ ول
oe ceri 227
ومد
As a Service
Pay-as-you-go or Subscription
Se eee eects
صفحه 9:
Advantages of
Cloud Computing
From a Developer’s Perspective:
Pay-as-you-go “utility computing”
0 ترا
0
On-demand resource allocation & release
Scalability
“ More on this later
0 ere (ere er
صفحه 10:
Risks of
Cloud Computing
Security
Who else has access to “your” resources ?
Recovery
۲ ۲۱۵۸ 7و 5و6
Provider lock-in
ce coe en Ok gt 90
صفحه 11:
Amazon’s Cloud Platform:
Amazon Web Services
Processors & Memory Network
“Elastic Compute Cloud Content Delivery
“EC2” Network CloudFront
Messaging تست
Simple Storage Service 0
“S3” 2205
| Elastic Block Store “EBS”
| SimpleDB database
3 ميق وا وا مرول Se eee
صفحه 12:
Java Technology Day — كن
ElasticFox Firefox plugin
صفحه 13:
Developing on
| NC Mid
Standard stuff: Challenges:
Language Scaling
Libraries
Communications
واع لمع 5 ماعلالا
Application Servers
Databases
mee for ee applications
1۱ ees te
صفحه 14:
Google’s Cloud Platform:
Google App Engine
Platform-as-a-Service ۷:
Language User Accounts
۲ اتهمع ما
1 ووز
Image Transformation
Storage
7 م,منوهاده »و ۲۴۸۵و وه ۰ ۵۵
واطم
عطوز ۲۵۳
| eee ete et et ۳
صفحه 15:
Ee
0
Java Technology Day
Google Dashboard هل
Google Administration Console
صفحه 16:
Developing on
Google’s Cloud ك0
۴۵5 ۰ Challenges:
Scaling Language
Libraries
Communications
Data Storage
mae for new, lighter-weight applications
SO eet ets
صفحه 17:
Application Development
Challenges Posed by the
Cloud
Deploying to the Cloud
Designing for Scalability
eon ie
Application Tier
ات۱۰
صفحه 18:
Application Development
Challenges Posed by the
Cloud
Deploying to the Cloud 9
Designing for Scalability
” ome
Application Tier
ات۱۰
صفحه 19:
ا ی
Deploying to the Cloud
laaS platforms
Mostly the same as traditional deployment
PaaS & SaaS platforms
Custom procedures
Custom configurations
تور ۹۲۵
“Google App Engine: Eclipse plug-in
eee en et et
صفحه 20:
4: ۵ Java Technology Day —
Deploy
صفحه 21:
Java Technology Day =: ی
0 لعدة صا نوا.. مرعل جلقدا متجع تفي see
صفحه 22:
الس مد
Designing the Application
Tier for Scalability
Me) Lena 4-1) ات ات۸۰۱۱ عناد عاقالا
Optimize database queries
Use in-memory caching
Parallelize operations
Use concurrent threads
“Use the Service Pools design pattern
eee eer et or 62
صفحه 23:
Parallelize with
Concurrent Threads
ارهز
Allow long-running tasks to proceed without
impacting performance
Java offers the java.util.concurrent
حتاف
“ Executor interface
ورین ات نا
eee en et et مه
صفحه 24:
java.util.concurrent
Example: In-Memory Cache
Existing implementations such as memcached
Cache shared by all application instances
Access is via the network
Application requests an Object from the cache
Time until response is received can vary
“ True of any network operation
Don’t let application code wait...
Se eee eects
000 0 0000
صفحه 25:
با
به Bee een x
Eee gee eee eee rar 0
mercy 2 o ا ا الل لل te
perform the rest of the request handling code here f= 77
چم 2
eee recs = عمو رقم 6ه عه 0]
ان ا 0 2
ce &
ey 50
1-2
yee اه
” as 62 2
pee enero 0
nes ere eerste 2
0 (Timeoutexception e) { رو أ
eer eae eee rune seme crs -
Sere reece ae) © 6ه
7 23
rr:
000 ene eee ner
eM ree re فد ل
صفحه 26:
Poe ee زماهمااه
1
0
cers ics
۱ tants
۵۲
5 0
7 000 ay
a|dwexy
eer ees ance)
ها هه عم ز0ن
: 0
00 = objectKey; 0
FutureTaskeT> getFroacachedperation = FutureTaskeT>( (iy ee
iene
ل ee
Cees)
0
4
7
۱ ant
00
لواو 1 5ه
لت 0
tee ere
صفحه 27:
Parallelize with Service Pools
لواسك ات سانا
Allow services to scale according to demand
“Scale up and down
سا
9 Response Queue 0 مما
eee tt er
صفحه 28:
Java Service Pool for
Amazon Web Services:
Open source 0 2.0
“ http://code.google.com/p/lifeguard/
۱
a 9
م Response Queue 7۷7 "
i
[1 a koe Eo
Onis
صفحه 29:
ام ava Technology ی
Lifeguard Framework
Framework provides:
Message handling
File handling
Scaling logic
Request SQS Queue
Ingestor
3 011۳
ار رس
<_ <
~~ per le a ۳
۷
5
1 en
ل
صفحه 30:
Lifeguard Framework
You provide:
Ingestor
عءألااع5
Pool Manager Configuration
2000 Request SQS Queue
ao} رس
<_ <
Bai :
~~ per le 1
۰
ae
10 eet ا
0
صفحه 31:
Pees eer a
RS CSE aa
ص0۳
eres
‘lane>Resizelnage</Nane>" +
pies ete ete nes
هر
و
0000 ac g
0111000 0
ere One
1 eee eee ree
إلنا013|01ا313 31/3 10353610
ل اا ال سس
صفحه 32:
١ |۱۳ معط
۳
و
| ee eae es
Bere ire eetscer str en aes
00
Precast creases
0
0
reenter 5
|
| ctr cry
0 te
إللا13|0نا3ناح لاد ©
دقاعت
eee ا ee a 88
صفحه 33:
۳
ee ore eres
ee eres
0 tem crs
pee eae eee sneaaies
See et ete
<RanpDownInterval>1</RanpDownInterval>
ee ens
pore wares
pee ere
Sree erro er ties
rene
This configuration defines the SLA for this
service
That's all there is to implement
The framework does all the rest os
ete ere ee ۱
sJaHeue| |OOg 9y} SunHyuoD
صفحه 34:
Service Pool
Pool of service instances dynamically scales with
load
Request SQS Queue
ری
۳ 2 ]
ro
>
os
~~ per le 1 F |
9
عل
ل ا ete
ار رس
٠
صفحه 35:
Service Pool
Pool of service instances dynamically scales with
load
هه سح
صفحه 36:
Service Pool
Multiple service pools scale independently
۲ ۳۳ 6
صفحه 37:
Service Pool
Workloads can follow different workflows
| Specify the Ingestor’s XML accordingly
Ea
at ال
صفحه 38:
aha 3 ای
Developing Cloud
Computing Applications with
Java
Q&A
ال 1
صفحه 39:
aha 3 ای
Developing Cloud
Computing Applications with
Java
تست سرت
shlomo.swidler@gmail.com
Thank you!
وس بو
Developing
Cloud
Computing
Applications
Shlomo Swidler
MyDrifts.com
withCTO,Java
shlomo.swidler@gmail.co
m
Developing Cloud
Computing Applications with
Java
Overview of Cloud Computing
Amazon’s Cloud Platform
Google’s Cloud Platform
Application Development Challenges Posed by
the Cloud…
… and Java Solutions
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
2
About Me
CTO & co-Founder
Music marketing on social networks
Patent-pending targeting technology
Java, MySQL, auto-scaling & cloud-based
Active in the Cloud Computing community
Open Cloud Computing Interface Working Group
(an Open Grid Forum initiative) participant
Contributor to Open Source cloud & Java projects
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
3
Cloud Computing Is…
A style of computing in which dynamically
scalable and often virtualized resources are
provided as a service over the Internet. Users
need not have knowledge of, expertise in, or
control over the technology infrastructure in
the “cloud” that supports them.
– Wikipedia
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
4
Cloud Computing Is…
A style of computing in which dynamically
scalable and often virtualized resources are
provided as a service over the Internet. Users
need not have knowledge of, expertise in, or
control over the technology infrastructure in
the “cloud” that supports them.
– Wikipedia
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
5
Cloud Computing Is…
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
6
Cloud Computing Is…
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
7
Cloud Computing Is…
Computing Resources
Infrastructure
Platform
Software
Processor
LAMP Stack
Email
Memory
JVM
CRM System
Storage
Python VM
ERP System
Network
MapReduce
SCM System
As a Service
22 June 2009
Pay-as-you-go or Subscription
Developing Cloud Computing Applications with Java
by Shlomo Swidler
8
Advantages of
Cloud Computing
From a Developer’s Perspective:
Pay-as-you-go “utility computing”
Saves time
Saves $$$
On-demand resource allocation & release
Scalability
22 June 2009
More on this later
Developing Cloud Computing Applications with Java
by Shlomo Swidler
9
Risks of
Cloud Computing
Security
Who else has access to “your” resources ?
Recovery
How easy is it ?
Provider lock-in
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
10
Amazon’s Cloud Platform:
Amazon Web Services
Infrastructure-as-a-Service
Processors & Memory
Elastic Compute Cloud
“EC2”
Storage
22 June 2009
Simple Storage Service
“S3”
Elastic Block Store “EBS”
SimpleDB database
Developing Cloud Computing Applications with Java
Network
Content Delivery
Network CloudFront
Messaging
Simple Queue Service
“SQS”
by Shlomo Swidler
11
Amazon Dashboard
ElasticFox Firefox plugin
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
12
Developing on
Amazon’s Cloud
Standard stuff:
Challenges:
Language
Scaling
Libraries
Communications
Web Servers
Application Servers
Databases
Suitable for existing applications
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
13
Google’s Cloud Platform:
Google App Engine
Platform-as-a-Service
Language
Python
Java
Storage
22 June 2009
User Accounts
Email
Image Transformation
Memcached
JDO or JPA or Datastore
APIs
Developing Cloud Computing Applications with Java
Cron jobs
by Shlomo Swidler
14
Google Dashboard
Google Administration Console
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
15
Developing on
Google’s Cloud
Easy stuff:
Challenges:
Scaling
Language
Libraries
Communications
Data Storage
Suitable for new, lighter-weight applications
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
16
Application Development
Challenges Posed by the
Cloud
Deploying to the Cloud
Designing for Scalability
Web Tier
Application Tier
22 June 2009
Database Tier
Developing Cloud Computing Applications with Java
by Shlomo Swidler
17
Application Development
Challenges Posed by the
Cloud
Deploying to the Cloud
Designing for Scalability
Web Tier
Application Tier
22 June 2009
Database Tier
Developing Cloud Computing Applications with Java
by Shlomo Swidler
18
Deploying to the Cloud
IaaS platforms
Mostly the same as traditional deployment
PaaS & SaaS platforms
Custom procedures
Custom configurations
Custom tools
22 June 2009
Google App Engine: Eclipse plug-in
Developing Cloud Computing Applications with Java
by Shlomo Swidler
19
Deploying an Application to
Google App Engine
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
20
Designing the Application
Tier for Scalability
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
21
Designing the Application
Tier for Scalability
Make sure your Storage Tier is optimized
Optimize database queries
Use in-memory caching
Parallelize operations
Use concurrent threads
Use the Service Pools design pattern
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
22
Parallelize with
Concurrent Threads
Motivation
Allow long-running tasks to proceed without
impacting performance
Java offers the java.util.concurrent
package
22 June 2009
Executor interface
Future<T> interface
Developing Cloud Computing Applications with Java
by Shlomo Swidler
23
java.util.concurrent
Example: In-Memory Cache
Existing implementations such as memcached
Cache shared by all application instances
Access is via the network
Application requests an Object from the cache
Time until response is received can vary
True of any network operation
Don’t let application code wait…
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
24
// perform the rest of the request handling code here
// then, at the end, get the user's last login date
Date lastLoginDate = null;
try {
lastLoginDate = lastLoginDateGetter.get(50, TimeUnit.MILLISECONDS);
} catch (InterruptedException e) {
// someone interrupted the FutureTask
} catch (ExecutionException e) {
// the FutureTask threw an exception
} catch (TimeoutException e) {
// the FutureTask didn't complete within the 50ms time limit
lastLoginDateGetter.cancel(false);
}
// return lastLoginDate to the presentation layer
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
java.util.concurrent
String memcachedKey = "userId:" + userId + ".lastLoginDate";
Future<Date> lastLoginDateGetter = MemcachedClient.get(memcachedKey, Date.class);
Example: In-Memory
Cache
String userId = "visitor01";
25
public class MemcachedClient {
private static Executor executor = Executors.newFixedThreadPool(1);
public static <T> Future<T> get(String objectKey, Class<T> type) {
final String objectKeyFinal = objectKey;
FutureTask<T> getFromCacheOperation = new FutureTask<T>(
new Callable<T>() {
public T call() {
Object networkResponse = requestObjectOverNetwork(objectKeyFinal);
return (T) networkResponse;
}
}
);
executor.execute(getFromCacheOperation);
return getFromCacheOperation;
}
private static Object requestObjectOverNetwork(String objectKey) {
// network stuff goes in here
}
java.util.concurrent
java.util.concurrent.Callable;
java.util.concurrent.Executor;
java.util.concurrent.Executors;
java.util.concurrent.Future;
java.util.concurrent.FutureTask;
Example: In-Memory
Cache
import
import
import
import
import
}
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
26
Parallelize with Service Pools
Motivation
Allow services to scale according to demand
Scale up and down
Request Queue
Response Queue
Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
27
Java Service Pool for
Amazon Web Services:
Open source Apache
License Version 2.0
Lifeguard
http://code.google.com/p/lifeguard/
Request Queue
Response Queue
Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
28
Lifeguard Framework
Framework provides:
Message handling
File handling
Scaling logic
Listener
Response SQS Queue
S3 Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
EC2 Instances
Ingestor
Request SQS Queue
Pool
Mgr
Config
29
Lifeguard Framework
You provide:
Ingestor
Service
Pool Manager Configuration
Listener
Response SQS Queue
S3 Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
EC2 Instances
Ingestor
Request SQS Queue
Pool
Mgr
Config
30
private static final String ResizeImageWorkflowXML =
"<Workflow>" +
"<Service>" +
"<Name>ResizeImage</Name>" +
"<WorkQueue>ResizeImage-input</WorkQueue>" +
"</Service>" +
"</Workflow>";
public ResizeImageIngestor() {
super(ResizeImageIngestorWorkflowXML);
}
public void ingest(File imageFile) {
super.ingest(Collections.singletonList(imageFile));
}
}
Ingestor
S3 Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
Implement the Ingestor
public class ResizeImageIngestor extends IngestorBase {
31
public ResizeImageService() {
super(ServiceConfigXML);
}
public List<File> executeService(File imageFile) {
Image origImage = new Image(imageFile);
File resizedImageFile = resizeImageToFile(origImage);
return Collections.singletonList(resizedImageFile);
}
}
S3 Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
Implement the Service
public class ResizeImageService extends AbstractBaseService {
private static final String ServiceConfigXML =
"<ServiceConfig>" +
"<ServiceName>ResizeImage</ServiceName>" +
"<WorkQueue>ResizeImage-input</WorkQueue>" +
"</ServiceConfig>";
32
<ServiceName>ResizeImage</ServiceName>
<VMImage>ami-39ba5df0</VMImage>
<WorkQueue>ResizeImage-input</WorkQueue>
<RampUpInterval>1</RampUpInterval>
<RampUpDelay>360</RampUpDelay>
<RampDownInterval>1</RampDownInterval>
<RampDownDelay>480</RampDownDelay>
<MinSize>0</MinSize>
<MaxSize>20</MaxSize>
<QueueSizeFactor>20000</QueueSizeFactor>
</ServicePool>
This configuration defines the SLA for this
service
That’s all there is to implement
The framework does all the rest
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
Pool
Mgr
Config
Configure the Pool Manager
<ServicePool>
33
Service Pool
Pool of service instances dynamically scales with
load
Listener
Response SQS Queue
S3 Storage
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
EC2 Instances
Ingestor
Request SQS Queue
Pool
Mgr
Config
34
Service Pool
Pool of service instances dynamically scales with
load
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
35
Service Pool
Multiple service pools scale independently
etc.
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
36
Service Pool
Workloads can follow different workflows
Specify the Ingestor’s XML accordingly
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
37
Developing Cloud
Computing Applications with
Java
Q&A
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
38
Developing Cloud
Computing Applications with
Java
Shlomo Swidler
shlomo.swidler@gmail.com
Thank you!
22 June 2009
Developing Cloud Computing Applications with Java
by Shlomo Swidler
39