building scalable websites for the cloud

55
1 Scalable Web Applications Reference Architectures and Best Practices Brian Adler, PS Architect

Upload: rightscale

Post on 27-Jun-2015

1.123 views

Category:

Technology


1 download

DESCRIPTION

Brian Adler, Solutions Consultant at RightScale, led this session at the RightScale User Conference 2010 in Santa Clara. Session Abstract: RightScale has extensive experience building scalable websites from the ground up. More importantly, we have invaluable experience rescuing companies who have come to us after their self-built websites failed to scale when hit with unexpected traffic levels. In this session, we'll review reference architectures and share best practices from proven, scalable websites deployed in the cloud. Additionally, we'll cover alternate methods for load balancing along with techniques to improve your web application's availability and reliability.

TRANSCRIPT

Page 1: Building Scalable Websites for the Cloud

1

Scalable Web ApplicationsReference Architectures and Best Practices

Brian Adler, PS Architect

Page 2: Building Scalable Websites for the Cloud

2

Scalable Web Application

Page 3: Building Scalable Websites for the Cloud

3

Scalable Web Application• What?

• An application built on an architecture that can adapt to changing conditions

Page 4: Building Scalable Websites for the Cloud

4

Scalable Web Application• What?

• An application layered on an architecture that can adapt to changing conditions

• Why?• Traffic and load patterns are unpredictable

• Viral or flash-mob events can result in very dynamic conditions

• Availability and Reliability• Application must be distributed to increase likelihood of end-user accessibility

• Overprovision• Under-utilized resources == wasted $$$

• Underprovision• Missed opportunities – users unable to access your site/product• Don’t be a victim of your own success

Page 5: Building Scalable Websites for the Cloud

5

This bed is too big. This bed is too small…

Page 6: Building Scalable Websites for the Cloud

6

Cloud Resource Model• Dynamically provision the resources you need to meet the

current demand• Demand goes up, resources are added• Demand goes down, resources are removed

• In true “utility computing” fashion, only pay for what you use, when you use it

Page 7: Building Scalable Websites for the Cloud

7

But this bed is just right

Page 8: Building Scalable Websites for the Cloud

8

Scalable Web Application• When?

• No time like the present

Page 9: Building Scalable Websites for the Cloud

9

Scalable Web Application• When?

• No time like the present

• How?• Stay tuned…

Page 10: Building Scalable Websites for the Cloud

10

Reference Architecture

Page 11: Building Scalable Websites for the Cloud

11

Load Balancing Tier

Page 12: Building Scalable Websites for the Cloud

12

Load Balancing• ELB or not ELB. That is the question.

• No SSL termination on the ELB (*)• Can load balance at the TCP level, but that eliminates sticky sessions for

secure connections (*)• (*) No longer the case as of mid-October 2010

• Can scale to handle large amounts of traffic, but slow to ramp-up• Only need one• (RightScale has a technical white paper on load balancing solutions that

can be provided if desired)

Page 13: Building Scalable Websites for the Cloud

13

Load Balancing• ELB or not ELB. That is the question.

• No SSL termination on the ELB (*)• Can load balance at the TCP level, but that eliminates sticky sessions for

secure connections (*)• (*) No longer the case as of mid-October 2010

• Can scale to handle large amounts of traffic, but slow to ramp-up• Only need one• (RightScale has a technical white paper on load balancing solutions that

can be provided if desired)

• HAProxy + Apache• Can handle SSL termination on the load balancer• Allows for sticky sessions with secure connections• Each instance can handle a specific amount of traffic with no ramp-up time• Each instance can only handle a specific amount of traffic• Addition of load balancers is possible, but requires DNS modifications

Page 14: Building Scalable Websites for the Cloud

14

Load Balancing• Load Balancer + Application server

• Possible, and good for test and dev• Not a best practice for a production environment

• Traffic spikes can cause instance to perform both load balancing and application functions…poorly

Page 15: Building Scalable Websites for the Cloud

15

Load Balancing• Load Balancer + Application server

• Possible, and good for test and dev• Not a best practice for a production environment

• Traffic spikes can cause instance to perform both load balancing and application functions…poorly

• Recommendation: Minimum of two load balancers• Each load balancer should be in a different availability zone (AZ) to

increase reliability and availability• RightScale testing has shown that m1.large is a good choice for load

balancers• Due to 100K-120K packet-per-second limit, larger instances do not provide much gain in

throughput• Roughly 5K responses/second can be handled by m1.large• With the 5K threshold in mind, select the number of load balancers required to handle your

peak traffic

Page 16: Building Scalable Websites for the Cloud

16

Application Server Tier• Puts the “scalable” in a scalable application• True autoscaling a must in any dynamic/unpredictable

environment

Page 17: Building Scalable Websites for the Cloud

17

Application Server Tier• Autoscaling

• Fully automated server launch based on autoscaling triggers• No manual intervention (can be challenging in certain environments, i.e.

Windows)• Download and install application code from common repository to ensure

identical configuration of all servers in the tier

Page 18: Building Scalable Websites for the Cloud

18

Application Server Tier• Autoscaling

• Fully automated server launch based on autoscaling triggers• No manual intervention (can be challenging in certain environments, i.e.

Windows)• Download and install application code from common repository to ensure

identical configuration of all servers in the tier

• Triggers• Common

• CPU idle• Free memory• System load

• Custom• Web server connections• Application-specific metrics

Page 19: Building Scalable Websites for the Cloud

19

Application Server Tier• When to scale?

• Conservatively. Both up and down

Page 20: Building Scalable Websites for the Cloud

20

Application Server Tier• When to scale?

• Conservatively. Both up and down

• Up• Allow adequate lead time for new servers to become operational• Before system is negatively impacted• Look for trends in activity and react early• Worst that can happen: Charged for an extra instance hour

Page 21: Building Scalable Websites for the Cloud

21

Application Server Tier• When to scale?

• Conservatively. Both up and down

• Up• Allow adequate lead time for new servers to become operational• Before system is negatively impacted• Look for trends in activity and react early• Worst that can happen: Charged for an extra instance hour

• Down• When system has been underutilized for a consistent, consecutive period

of time• Scale down fewer servers than in a scale-up event• Again, only downside is an extra hour of instance charge• Better safe than sorry

Page 22: Building Scalable Websites for the Cloud

22

Application Server Tier• Array considerations

Page 23: Building Scalable Websites for the Cloud

23

Application Server Tier• Array considerations

• Weight the array across all availability zones (not regions)• Increases reliability of application• NOTE: Traffic within an AZ on private IPs is free. Traffic between AZs incurs a per-gigabyte

charge• Traffic between regions is charged at public Internet rates

Page 24: Building Scalable Websites for the Cloud

24

Application Server Tier• Array considerations

• Weight the array across all availability zones (not regions)• Increases reliability of application• NOTE: Traffic within an AZ on private IPs is free. Traffic between AZs incurs a per-gigabyte

charge• Traffic between regions is charged at public Internet rates

• Set minimums and maximums appropriately• Minimum can assist in cost savings in times of low usage• Maximum can limit overall cost exposure

Page 25: Building Scalable Websites for the Cloud

25

Application Server Tier• Array considerations

• Weight the array across all availability zones (not regions)• Increases reliability of application• NOTE: Traffic within an AZ on private IPs is free. Traffic between AZs incurs a per-gigabyte

charge• Traffic between regions is charged at public Internet rates

• Set minimums and maximums appropriately• Minimum can assist in cost savings in times of low usage• Maximum can limit overall cost exposure

• Instance size• m1.large is typically a good choice for array-based servers in a production environment

• m1.smalls (and even micro instances) can be used in test and development environments

• Every application is different, so run load tests and benchmarks to find the optimal solution for your environment

Page 26: Building Scalable Websites for the Cloud

26

Application Server Tier• Array considerations

• Weight the array across all availability zones (not regions)• Increases reliability of application• NOTE: Traffic within an AZ on private IPs is free. Traffic between AZs incurs a per-gigabyte

charge• Traffic between regions is charged at public Internet rates

• Set minimums and maximums appropriately• Minimum can assist in cost savings in times of low usage• Maximum can limit overall cost exposure

• Instance size• m1.large is typically a good choice for array-based servers in a production environment

• m1.smalls (and even micro instances) can be used in test and development environments

• Every application is different, so run load tests and benchmarks to find the optimal solution for your environment

• Code Deployment• Updated code can be pushed to all servers in an array via a single click of a button

Page 27: Building Scalable Websites for the Cloud

27

Caching Tier• Caching can dramatically decrease the load on the database

• Particularly in read-intensive applications

• Costs of caching• Application complexity/modification• Additional instance hours to support the cache

Page 28: Building Scalable Websites for the Cloud

28

Caching Tier• Best practice is to have a separate, dedicated caching tier

• Caching can be implemented on each application server• Prevents the use of a distributed cache• Local cache should only be used by the co-resident application server

• Application complexities• Database performance degradation

Page 29: Building Scalable Websites for the Cloud

29

Caching Tier• Best practice is to have a separate, dedicated caching tier

• Caching can be implemented on each application server• Prevents the use of a distributed cache• Local cache should only be used by the co-resident application server

• Application complexities• Database performance degradation

• Instance Size and Count• Determine memory caching footprint

• Select instance size and count that spreads the load over several servers• Prevents loss of entire cache if a single instance fails• Distribute caching servers across AZs for reliability• Overprovision if possible

• Provide capacity for system to grow to fully utilize cache (budget permitting)

Page 30: Building Scalable Websites for the Cloud

30

Caching Tier• Best practice is to have a separate, dedicated caching tier

• Caching can be implemented on each application server• Prevents the use of a distributed cache• Local cache should only be used by the co-resident application server

• Application complexities• Database performance degradation

• Instance Size and Count• Determine memory caching footprint

• Select instance size and count that spreads the load over several servers• Prevents loss of entire cache if a single instance fails• Distribute caching servers across AZs for reliability• Overprovision if possible

• Provide capacity for system to grow to fully utilize cache (budget permitting)

• Manually scaling caching servers is possible but non-trivial• Involves application and database performance degradation• Time To Lives (TTLs)

• Always set to expire

Page 31: Building Scalable Websites for the Cloud

31

Caching Tier• Write-intensive applications

• Don’t see as large a performance gain as read-intensive apps• Memcached can be modified to perform lazy writes of data objects to the

database• Data can be lost in case of caching server crash• Not a recommended best practice, but can be (and has been) done• Tradeoff of performance versus end-user experience

Page 32: Building Scalable Websites for the Cloud

32

Caching Tier• Write-intensive applications

• Don’t see as large a performance gain as read-intensive apps• Memcached can be modified to perform lazy writes of data objects to the

database• Data can be lost in case of caching server crash• Not a recommended best practice, but can be (and has been) done• Tradeoff of performance versus end-user experience

• Third-party providers• Vendor solutions exist that allow dynamic memcached scaling

Page 33: Building Scalable Websites for the Cloud

33

Database Tier• Numerous database architecture options exist• No “one size fits all” solution, so testing and benchmarking are

critical to determine proper configuration

Page 34: Building Scalable Websites for the Cloud

34

Database Tier• Masters and Slave(s)

• Multiple Slaves if budget permits• Distribute Master and Slave(s) across AZs• Always use same instance size for Master and Slaves

Page 35: Building Scalable Websites for the Cloud

35

Database Tier• Masters and Slave(s)

• Multiple Slaves if budget permits• Distribute Master and Slave(s) across AZs• Always use same instance size for Master and Slaves

• Data Storage• EBS volumes for data store• Never use ephemeral storage for persistent data• Back up Master and Slaves frequently• Upload snapshots to S3 or some other persistent, redundant storage

Page 36: Building Scalable Websites for the Cloud

36

Database Tier• Masters and Slave(s)

• Multiple Slaves if budget permits• Distribute Master and Slave(s) across AZs• Always use same instance size for Master and Slaves

• Data Storage• EBS volumes for data store• Never use ephemeral storage for persistent data• Back up Master and Slaves frequently• Upload snapshots to S3 or some other persistent, redundant storage

• Instance Size• Varies greatly based on the nature of the application and site traffic• Load testing and benchmarking can assist in identifying a reasonable

initial size

Page 37: Building Scalable Websites for the Cloud

37

Database Scaling

Page 38: Building Scalable Websites for the Cloud

38

Database Scaling• Vertical

• “Grow” or “shrink” a database server from one instance size to another

Page 39: Building Scalable Websites for the Cloud

39

Database Scaling• Vertical

• “Grow” or “shrink” a database server from one instance size to another

• Horizontal• Add additional servers to spread the database load

Page 40: Building Scalable Websites for the Cloud

40

Database Vertical/Horizontal Scaling

RightScale Makes Vertical or Horizontal Scaling Easier

SmallType A

SmallType B

SmallType B

SmallType B

SmallType A

SmallType A

LargeType A

LargeType B

HorizontalScaling

VerticalScaling

More servers of same types

Same quantity of larger servers

Page 41: Building Scalable Websites for the Cloud

41

Horizontal Database Scaling• Addition of read Slaves

• Effective for read-intensive applications• Only writes need to access the master• Replication lag to slaves must be considered

Page 42: Building Scalable Websites for the Cloud

42

Horizontal Database Scaling• Addition of read Slaves

• Effective for read-intensive applications• Only writes need to access the master• Replication lag to slaves must be considered

• Effective mechanism is to use MySQL Proxy

Page 43: Building Scalable Websites for the Cloud

43

Horizontal Database Scaling• Sharding

• Concept is to partition the database into distinct, non-overlapping pieces• “Horizontal slicing” of the database tables into groups of rows• Forethought required in setting up shards since cross-shard joins are

resource intensive

Page 44: Building Scalable Websites for the Cloud

44

Horizontal Database Scaling

Before Sharding

Page 45: Building Scalable Websites for the Cloud

45

Horizontal Database Scaling

After Sharding

Page 46: Building Scalable Websites for the Cloud

46

Horizontal Database Scaling• Master-Master

• Two (or more) Master DBs• Any Master can modify any database object• Replication lag can result in database inconsistencies• Poorly-designed applications can cause data object collisions and leave

databases in indeterminate state• Not a recommended best practice, nor supported by RightScale

Page 47: Building Scalable Websites for the Cloud

47

Horizontal Database Scaling• NoSQL solutions

• Many options exist – SimpleDB, Cassandra, Membase, CouchDB, MongoDB, Riak, etc.

• Basically a Key/Value store• No complex operations between data objects (no relational operations)• Multiple nodes can be used to implement a distributed data store

• Coordinated backup and recovery can be challenging

• Some RightScale customers are beginning to use some NoSQL solutions in specific use cases.

Page 48: Building Scalable Websites for the Cloud

48

So What’s Best?

Page 49: Building Scalable Websites for the Cloud

49

So What’s Best?• As is typical in many technology discussions…

Page 50: Building Scalable Websites for the Cloud

50

So What’s Best?• As is typical in many technology discussions…

“It depends”

Page 51: Building Scalable Websites for the Cloud

51

So What’s Best?• As is typical in many technology discussions…

• Many scalable environments share some common underlying architecture concepts

“It depends”

Page 52: Building Scalable Websites for the Cloud

52

So What’s Best?• As is typical in many technology discussions…

• Many scalable environments share some common underlying architecture concepts

• Every application is different. As such, there is no “one size fits all”

“It depends”

Page 53: Building Scalable Websites for the Cloud

53

So What’s Best?• As is typical in many technology discussions…

• Many scalable environments share some common underlying architecture concepts

• Every application is different. As such, there is no “one size fits all”

• Components of a reference architecture such as this can be used as a starting point, with tweaks and modifications made per the unique characteristics of the application itself, or the load and traffic patterns it experiences

“It depends”

Page 54: Building Scalable Websites for the Cloud

54

Scalable Web Applications

Q&ARightScale.com/Conference

(Presentations available next week)

[email protected]

RightScale.com/whitepapers

Page 55: Building Scalable Websites for the Cloud

55