CloudStack Hyderabad Meetup: Migrating applications to IaaS clouds
1 like4,491 views
The document discusses migrating applications to Infrastructure as a Service (IaaS) clouds. It provides guidance on identifying candidate applications for migration based on varying resource needs. It also covers choosing an appropriate IaaS provider based on features, developing a migration strategy including rearchitecting applications for scalability, and presents a case study of migrating a social commerce platform to the cloud.
![About
me
Chirag
Jog
• Co-‐founder
and
CTO,
Clogeny
Technologies
• Experience
in
building
some
cuYng
edge
products
in
the
cloud.
• Extensive
experGse
across
IaaS,
PaaS
&
SaaS
–
20+
cloud
pla]orms
• Deep
experGse
in
Apache
Cloudstack
Pla]orm
• Several
deployments
for
marquee
customers
©
2012
Clogeny
Technologies
h8p://www.clogeny.com](https://image.slidesharecdn.com/cloudstack-meetup-migrating-to-iaas-chirag-121102103001-phpapp01/85/CloudStack-Hyderabad-Meetup-Migrating-applications-to-IaaS-clouds-2-320.jpg)
![A
case-‐study:
Social
commerce
pla]orm
The
customer
:
US-‐based
VC-‐funded
startup
in
the
social
commerce
and
social
shopping
recommendaGon
space.
Challenges:
" Hosted
on
non-‐ElasGc
infrastructure
–
shared
hosGng
" 3rd
party
plugin
in
criGcal
path
–
Highly
Available
" 30-‐40x
increase
in
traffic
during
holiday
season
–
ability
to
scale-‐up/down
based
on
load
" AutomaGon
and
ease-‐of-‐use
" Reduce
TCO
©
2012
Clogeny
Technologies
h8p://www.clogeny.com](https://image.slidesharecdn.com/cloudstack-meetup-migrating-to-iaas-chirag-121102103001-phpapp01/85/CloudStack-Hyderabad-Meetup-Migrating-applications-to-IaaS-clouds-9-320.jpg)
![A
case-‐study:
Social
e-‐commerce
pla]orm
FaceBook
AnalyGcs
Server
Integra;on
Redis
Server
Tornado
WebServer
Redis
Node
Feed
User
Behavior
to
•
Mobile
Apps
Analy;cs
server
• Tablet
Apps
Nginx
FrontEnd
Webserver
Schedule
workers
to
RabbitMQ
Message
Broker
Local
Redis
Cache
perform
• Desktop
Browser
tasks
Pylons
Web
framework
Celery
Worker
MongoDB
User
Interface
StaAc
Content
Javascript/CSS
Serve
Sta;c
Store
all
Content
vendor,
through
a
customer
and
Apache
Webserver
separate
server
10
behavioral
data
in
MongoDB
3rd
Party
Monitoring
(Server
Density)
Exis;ng
Infrastructure
©
2012
Clogeny
Technologies
h8p://www.clogeny.com](https://image.slidesharecdn.com/cloudstack-meetup-migrating-to-iaas-chirag-121102103001-phpapp01/85/CloudStack-Hyderabad-Meetup-Migrating-applications-to-IaaS-clouds-10-320.jpg)
![A
case-‐study:
VMWare
to
Cloudstack
MigraGon
" The
customer
is
a
leading
managed
soluGon
provider
in
North
America.
" Challenges:
• Migrate
100+
VMware
based
VMs
to
Citrix
CloudPla]orm
3.0.5
• Design
and
develop
a
workflow
to
scan
exisGng
VMWare
vSphere
inventory,
create
images(OVAs)
and
import
into
Cloudstack
as
per
accounts.
• MulG-‐volume
OVA
import
into
Cloudstack
• Restoring
the
associaGon
between
public
IP
address
&
Virtual
Machines
©
2012
Clogeny
Technologies
h8p://www.clogeny.com](https://image.slidesharecdn.com/cloudstack-meetup-migrating-to-iaas-chirag-121102103001-phpapp01/85/CloudStack-Hyderabad-Meetup-Migrating-applications-to-IaaS-clouds-18-320.jpg)
Ad
More Related Content
What's hot (20)
Viewers also liked (20)
Ad
Similar to CloudStack Hyderabad Meetup: Migrating applications to IaaS clouds (20)
Ad
More from CloudStack - Open Source Cloud Computing Project (11)
CloudStack Hyderabad Meetup: Migrating applications to IaaS clouds
- 1. InnovaGon → ExecuGon → SoluGon → Delivered Cloud MigraGng ApplicaGons to IaaS Clouds Computing -‐ Chirag Jog Enterprise DevOps Applica2ons Public, Private, PaaS, SaaS Datacenter US: +1 408 556 9645 Management India: +91 20 661 43 400 Web: h8p://www.clogeny.com Email: [email protected] Clogeny ConfidenGal
- 2. About me Chirag Jog • Co-‐founder and CTO, Clogeny Technologies • Experience in building some cuYng edge products in the cloud. • Extensive experGse across IaaS, PaaS & SaaS – 20+ cloud pla]orms • Deep experGse in Apache Cloudstack Pla]orm • Several deployments for marquee customers © 2012 Clogeny Technologies h8p://www.clogeny.com
- 3. Agenda " IdenGfying the Candidates for the Cloud " Choosing the right IaaS cloud " MigraGon strategy " Case Studies " Conclusion © 2012 Clogeny Technologies h8p://www.clogeny.com
- 4. IdenGfying the Candidates " Does you applicaGon have varying degree of demands? • Need more horsepower at specific Gmes/seasons " Do you expect your compute, storage and network requirements to grow over Gme? • Avoid upfront costs. • Scale-‐up when required. " Is your tesGng and QA acGvity restricted to the hardware you can afford? • Provision addiGonal hardware only during QA/release cycle " Support, Licensing costs © 2012 Clogeny Technologies h8p://www.clogeny.com
- 5. Criteria to choose the right IaaS Cloud " Service Level Agreements (SLAs) " ElasGcity for Compute and Storage resources " Networking and Security features/services " Virtual Machines and Data Backups and Restores " Availability Regions and Zones " DevOps, AutomaGon " Monitoring " Image Management " AddiGonal services – Database as a service, Object storage, load balancers © 2012 Clogeny Technologies h8p://www.clogeny.com
- 6. Public and Private Cloud Providers " Leaders • Amazon Web Services, Terremark, Savvis " Challengers • Bluelock, Gogrid, Joyent " Cloudstack-‐based • Ninefold, TATA Instacompute, Instance.com " Private Clouds • Cloudstack, Eucalyptus, VMWare vCloud Director and Openstack © 2012 Clogeny Technologies h8p://www.clogeny.com
- 7. MigraGon Strategy Investigation & Automation & Final Setup Replication Phase Optimization Phase Phase Deployment Cloud Cluster & Cloud Asset Production Planning Architecture Performance Setup Monitoring & Analysis Automation Review current and Arrive at the right P e r f o r m a n c e Create the components Standard and custom f u t u r e b u s i n e s s topology that suits engineering, clustering, and con>igurations for built reports, metrics r e q u i r e m e n t s a n d customer needs, plan l o a d b a l a n c i n g , y o u r c l o u d a s s e t s and parameters on develop relevant plans s t o r a g e , s e c u r i t y , s h a r d i n g , c a c h i n g , i n c l u d i n g i m a g e s , resource utilization scripts, automation, for cloud adoption production automation, managing networks and and demand patterns m o n i t o r i n g a n d including choice of user monitoring tools storage performance plus load balancer for m a n a g e m e n t t o o l s . c l o u d v e n d o r a n d for analysis of resource Con>igure HA & auto-‐ your services on the platform usage scaling cloud © 2012 Clogeny Technologies h8p://www.clogeny.com
- 8. Re-‐architecGng for the cloud " Scalability: Horizontal, VerGcal, Geographical " Loose Coupling " Monitoring and Logging " Automate and operaGonalize as much as possible " Design for Failure © 2012 Clogeny Technologies h8p://www.clogeny.com
- 9. A case-‐study: Social commerce pla]orm The customer : US-‐based VC-‐funded startup in the social commerce and social shopping recommendaGon space. Challenges: " Hosted on non-‐ElasGc infrastructure – shared hosGng " 3rd party plugin in criGcal path – Highly Available " 30-‐40x increase in traffic during holiday season – ability to scale-‐up/down based on load " AutomaGon and ease-‐of-‐use " Reduce TCO © 2012 Clogeny Technologies h8p://www.clogeny.com
- 10. A case-‐study: Social e-‐commerce pla]orm FaceBook AnalyGcs Server Integra;on Redis Server Tornado WebServer Redis Node Feed User Behavior to • Mobile Apps Analy;cs server • Tablet Apps Nginx FrontEnd Webserver Schedule workers to RabbitMQ Message Broker Local Redis Cache perform • Desktop Browser tasks Pylons Web framework Celery Worker MongoDB User Interface StaAc Content Javascript/CSS Serve Sta;c Store all Content vendor, through a customer and Apache Webserver separate server 10 behavioral data in MongoDB 3rd Party Monitoring (Server Density) Exis;ng Infrastructure © 2012 Clogeny Technologies h8p://www.clogeny.com
- 11. MigraGon Steps – InvesGgaGon Phase • Understanding the exisGng architecture. • IdenGfy horizontal scalable components. • UGlize Cloudstack’s features most effecGvely. • Right-‐sizing of the servers: § Web servers : Mostly request forwarders § Celery workers: CPU Intensive § Datastores : Memory and CPU Intensive • Storage configuraGon for Volumes: § RAID configuraGon § Filesystem type § No. of volumes © 2012 Clogeny Technologies h8p://www.clogeny.com
- 12. MigraGon Steps – ReplicaGon Phase " Setup the servers and Volumes as per invesGgaGon " Setup the webserver and workers " Database replicaGon • MongoDB node set up as slaves in replicaGon mode. • Redis node are set up as slaves in master-‐slave mode. • Ensures all the data – exisGng and live data is replicated automaGcally. " FuncGonal and IntegraGon tesGng to ensure the applicaGon works as-‐is © 2012 Clogeny Technologies h8p://www.clogeny.com
- 13. MigraGon Steps – OpGmizaGon Phase " Database Layer • Convert MongoDB from single-‐node to mulG-‐node replica-‐ set for equal load distribuGon • Built-‐in facility for fail-‐over • Setup Redis as a master-‐slave with manual failover " ApplicaGon Server Layer • Move all the resources – message broker (RabbitMQ) and workers (Celery) to separate nodes. • Ability to scale to the workers (Celery) based on load " Web Server • Single point of failure • Leverage Cloudstack’s Load Balancer capabiliGes © 2012 Clogeny Technologies h8p://www.clogeny.com
- 14. MigraGon Steps – OpGmizaGon Phase " Monitoring and Alerts: • Ability to monitor all the components • track a variety of metrics to pro-‐acGvely idenGfy issues in producGon setup • Automated Email/SMS alerts or Mobile Apps for noGficaGons • SoluGons : Nagios, Hyperic, Cloudkick, Server Density • Metrics to be tracked : § process availability § server availability § free/used disk space § free/used memory § free/used swap © 2012 Clogeny Technologies h8p://www.clogeny.com
- 15. MigraGon Steps – AutomaGon Phase " Servers: Automated Provisioning, ConfiguraGon and Backup. " Data-‐stores : Automated backups with scheduling " Self-‐managing Virtual Machines " Tools: Chef, Puppet, CfEngine, Fabric © 2012 Clogeny Technologies h8p://www.clogeny.com
- 16. AutomaGon scripts " Provision all the type of servers, configure appropriate services and storage " Setup a instance by pulling the latest code " For Webservers, the tool configures the instance and a8aches it to an Load Balancer " For MongoDB/Redis: automaGcally join the cluster or create a new one. " Data Backups: • Automate the backups of the data stores arer quiescing the datastore. • Runs arer a configurable period of Gme. • Manage exisGng snapshots and delete really old snapshots © 2012 Clogeny Technologies h8p://www.clogeny.com
- 17. Cloud-‐based Setup Send Emails Regular backups of (for Alerts) Volumes using snapshots FaceBook IntegraGon Monitoring AnalyGcs Server & Alerts Redis Server Store User Master –Slave feeds Tornado Mode WebServer MongoDB Feed User Replica Set Behavior to Mode Analy;cs server Redis Cluster • Mobile Apps Nginx FrontEnd • Tablet Apps Webserver RabbitMQ Load Local Redis Cache • Desktop Balancer Message Pylons Web Schedule Broker framework workers to perform tasks MapRed uce User Interface Celery StaAc Content Worker Workers Javascript/CSS Serve Sta;c Load Content through a 17 Balancer Apache separate server Webserver Worker Role Cloudstack-‐based Infrastructure © 2012 Clogeny Technologies h8p://www.clogeny.com
- 18. A case-‐study: VMWare to Cloudstack MigraGon " The customer is a leading managed soluGon provider in North America. " Challenges: • Migrate 100+ VMware based VMs to Citrix CloudPla]orm 3.0.5 • Design and develop a workflow to scan exisGng VMWare vSphere inventory, create images(OVAs) and import into Cloudstack as per accounts. • MulG-‐volume OVA import into Cloudstack • Restoring the associaGon between public IP address & Virtual Machines © 2012 Clogeny Technologies h8p://www.clogeny.com
- 19. A case-‐study: VMWare to Cloudstack MigraGon " Key Features: • Developed a inventory scan engine using vSphere SOAP APIs to generate a spec of the complex hierarchies of datacenters, folders, hosts and clusters. • Developed a mulG-‐step workflow to: § scan a virtual machine hardware configuraGon and disk layout § generate a snapshot (VMDK) of the virtual machine § extract per-‐volumes OVAs as per Cloudstack’s expectaGons from the snapshot. • Leverage Cloudstack’s Lazy Assignment of IP Addresses to allocate a specific IP address for an account.* *Don’t try this at home. We modified the Cloudstack database directly © 2012 Clogeny Technologies h8p://www.clogeny.com
- 20. Thanks ! Contact: [email protected] © 2012 Clogeny Technologies h8p://www.clogeny.com