1.
Salesforce’s Multitenant
Architecture
​ Doug Merrett
​ Principal Enterprise Architect – Asia Pacific
​ dmerrett@salesforce.com
How we do the magic we do…

2.
Safe harbor statement under the Private Securities Litigation Reform Act of 1995:
This presentation may contain forward-looking statements that involve risks, uncertainties, and assumptions. If any such uncertainties
materialize or if any of the assumptions proves incorrect, the results of salesforce.com, inc. could differ materially from the results expressed
or implied by the forward-looking statements we make. All statements other than statements of historical fact could be deemed forward-
looking, including any projections of product or service availability, subscriber growth, earnings, revenues, or other financial items and any
statements regarding strategies or plans of management for future operations, statements of belief, any statements concerning new,
planned, or upgraded services or technology developments and customer contracts or use of our services.
The risks and uncertainties referred to above include – but are not limited to – risks associated with developing and delivering new
functionality for our service, new products and services, our new business model, our past operating losses, possible fluctuations in our
operating results and rate of growth, interruptions or delays in our Web hosting, breach of our security measures, the outcome of any
litigation, risks associated with completed and any possible mergers and acquisitions, the immature market in which we operate, our
relatively limited operating history, our ability to expand, retain, and motivate our employees and manage our growth, new releases of our
service and successful customer deployment, our limited history reselling non-salesforce.com products, and utilization and selling to larger
enterprise customers. Further information on potential factors that could affect the financial results of salesforce.com, inc. is included in
our annual report on Form 10-K for the most recent fiscal year and in our quarterly report on Form 10-Q for the most recent fiscal quarter.
These documents and others containing important disclosures are available on the SEC Filings section of the Investor Information section
of our Web site.
Any unreleased services or features referenced in this or other presentations, press releases or public statements are not currently available
and may not be delivered on time or at all. Customers who purchase our services should make the purchase decisions based upon features
that are currently available. Salesforce.com, inc. assumes no obligation and does not intend to update these forward-looking statements.
Safe Harbor

3.
Shared services across applications
The Customer Success Platform
APIs
2,700+ Partner Apps
Open Ecosystem
Workflow
Data &
Objects Identity
Fast App Dev & Customization
AnalyticsCollaborationMobile UI
Scalable Metadata Platform
Complete CRM
Trusted Multitenant Cloud
AnalyticsCommunityMarketingServiceSales Apps

4.
Multitenancy

5.
One Cloud with Many Customers
Shared Elastic Services
One Primary Data Store per Production Instance
8K+ Customers per Instance
50+ Production Instances
All data segregated by customer
All operations include Org ID
Disaster Recovery
Per Org encryption keys

6.
What is in an Instance
Pivot
tables
Data tables
Metadata
tables
Shared Database Metadata Cache
Bulk data
processing
Multitenant aware
Query optimizer
Runtime App
Generator
Full text search
engine
Virtual Application Components
Common
Application
Screens
Tenant
Specific
Screens
Objects
(Tables)

7.
What Multitenancy means for Salesforce R&D
No Legacy Teams
Bugs fixed for everyone
One Version

8.
What Multitenancy means for Salesforce R&D
No Legacy Teams
Bugs fixed for everyone
One Version
260K+ of our Tests
Run your tests as well
Automation

9.
What Multitenancy means for Salesforce R&D
No Legacy Teams
Bugs fixed for everyone
One Version
Instance Architecture
260K+ of our Tests
Run your tests as well
Automation
Staggered Releases
Scalability across all sizes

10.
What Multitenancy means for Salesforce R&D
Three major releases per year
Bug fixes every week
Predictability
No Legacy Teams
Bugs fixed for everyone
One Version
Instance Architecture
260K+ of our Tests
Run your tests as well
Automation
Staggered Releases
Scalability across all sizes

11.
What Multitenancy means for Salesforce R&D
Three major releases per year
Bug fixes every week
Predictability
No Legacy Teams
Bugs fixed for everyone
One Version
Instance Architecture
260K+ of our Tests
Run your tests as well
Automation
Staggered Releases
Scalability across all sizes

12.
​ Stateless Appservers
​ Database system of record
​ No Database Definition Language (DDL) at Runtime
​ All tables partitioned by OrgId
​ Smart Primary Keys, Polymorphic Foreign Keys
​ Creative de-normalization and pivoting
​ Use every RDBMS feature & optimization
Key Architectural Principles

13.
Metadata, data, and pivot table structures store data
corresponding to virtual data structures

14.
The Objects table stores metadata about custom objects
(tables)‫‏‬

15.
The Fields table stores metadata about custom fields
(columns)‫‏‬

16.
The Data heap table stores all structured data corresponding
to custom objects

17.
A single slot can store various types of data that originate
from different objects

18.
The Indexes pivot table manages tenant-specific selective
indexes

19.
The UniqueFields pivot table facilitates uniqueness for
custom fields

20.
The Relationships pivot table facilitates referential integrity
and optimizes joins

21.
​ Tables hash partitioned by OrgId
​ Separate connection pools point to physical hosts
​ App tier is also dynamically partitioned by OrgId
​ Distributed metadata cache with transactional invalidation
All data & metadata structures are partitioned to improve
performance and manageability

22.
§ Native Declarative features
§ Bulk Processing
§ The Recycle Bin
§ Full Text Search
§ Smart Bulk Data Manipulation Language (DML)
§ Web Services APIs
Application Framework: a whole lot for free

23.
Force.com’s native Application Framework provides
declarative development, no coding

24.
Validation rules and simple formulas:
Business analysts can “code” these

25.
Not so simple: Rollup-summary fields provide for easy cross-
object summaries

26.
Force.com’s bulk processing optimizations reduce overhead
for data loads

27.
​ Examples:
§ Sort all records by primary key before attempting DML
§ Operate on tables in deterministic order
§ Slot reallocation for field datatype change
§ Deferred calculation for new rollup-summary field
§ Background processing of mass changes
Data definition processing is optimized to avoid performance
hits or concurrency limits

28.
The Recycle Bin: Smart Undeletes
Restore
§ Individual object instances
(records)
§ Related object instances (parent/
child records)
§ Entire fields and objects
(dropped columns and tables)

29.
Secondary Instance
Multitenant Search, anything but simple
Index Backup
Replication
Primary Instance

30.
Multitenancy delivers Blazing Performance
Transactions Per Quarter
234B Transactions in Q2FY16
79% YoY Growth
Average Page Time
208ms Latency in Q2FY16
4% YoY Improvement
API
Web

31.
​ Consistent SQL generation across the application
​ Deep awareness of pivot table structure
• Flex schema does impose a cost
​ Tenant, user, object, fields statistics are crucial
​ No runaway queries allowed
​ Deep integration with the sharing model
Multitenant Query Optimization Principles

32.
Multitenant Query Optimizer
Check user
visibility
Check filer
selectivity
Dynamically
write query
based on pre-
queries
Run Pre-Queries
Execute
optimized
query
user visibility = number of rows user can access
filter selectivity = index corresponding to filter column
Search originates
from API or global
search
return results

33.
The optimizer considers pre-query selectivity measurements
when writing a query
Pre-Query
Selectivity
Measurements
… use of index related to filter.HighHigh
… ordered hash join; drive using Data table.LowHigh
… use of index related to filter.HighLow
… nested loops join; drive using view of rows that the user can see.LowLow
Write final database access query, forcing …
FilterUser

34.
Highly Scalable Multitenant
Architecture

35.
Service isolation and redundant design ensures availability
Highly Scalable Multitenant Architecture
• Instances
– Consistent deployment size
– Easy to scale
– Repeatable
• Instance Groups
– Shared service isolation
within a Data Center
• Data Centers
– Consistent for Production
and Disaster Recovery
– Shared services across data
centers
Data Centre
Shared Services
Instance Group
Shared
Services
Sandbox DR
Instances
Sandbox
Instances
Core DR
Instances
Core
Instances
Instance Group
Shared
Services
Sandbox DR
Instances
Sandbox
Instances
Core DR
Instances
Core
Instances
Data Centre
Shared Services
Instance Group
Shared
Services
Sandbox DR
Instances
Sandbox
Instances
Core DR
Instances
Core
Instances
Instance Group
Shared
Services
Sandbox DR
Instances
Sandbox
Instances
Core DR
Instances
Core
Instances

36.
​ Production cluster capacity
• 7+1 (active + spare) nodes
• Active Data Guard to local standby
(2 node cluster) for fast data
recovery
​ Data replication to secondary
instance
• Encrypted block-level replication
• Enables site switching for disaster
recovery
Redundant Site Design with 4 Online Copies of Data
Today: Reliable Core by Design
Encrypted
Async
ReplicationData Guard
Replication
Primary Instance
Application
Servers
Production
DB Cluster
Standby
DB Cluster
Data Guard
Replication
Application
Servers
Production
DB Cluster
Standby
DB Cluster
Secondary Instance

37.
​ 6+2 Production RAC Cluster
• 2 spare nodes provides additional
DB cluster resilience
​ Data Guard to remote site
• Hardens remote replication
• Enables faster site switching
• Shorter planned and unplanned
maintenance windows
Faster Site Switches
Tomorrow: Enhancing Core Reliability
Encrypted
Async DB
Replication
Data Guard
Replication
Primary Instance
Application
Servers
Production
DB Cluster
Standby
DB Cluster
Data Guard
Replication
Application
Servers
Production
DB Cluster
Standby
DB Cluster
Secondary Instance

38.
4 key methods to enable performance at scale
Adaptive Capacity Planning
​ Horizontally
​ Scaling
​ Scale by adding
additional servers or
storage for each
application
​ Scale by using more
powerful servers
• More CPU cores
• Larger memory
​ Continuous Salesforce
application and
database optimizations
​ Regular customer and
partner application
tuning to improve
individual org
performance
​ Vertically
​ Scaling
​ Application and DB
​ Tuning
​ Customer Application
​ Tuning
45% Reduction in
Latency
NA7
Aug 20
NA7
Aug 27
360ms
250ms
140+ Hours
Loading Time
21 Hours
Loading Time
85% Reduction in
Load Time
1 2 3 4
Automatically Done For Customers
48c
48c
64c
64c

39.
Force.com Extras

40.
Apex: Force.com’s procedural frontier
Integer NUM = 10;
Account [] accs;
// Clean up old data
accs = [select id from account where name like 'test%'];
delete accs;
commit;
accs = new Account [NUM];
for (Integer i = 0; i < NUM; i++)
{
accs [i] = new Account (name='test ' + i, outstandingshares__c=i);
}
insert accs;
Contact [] cons = new Contact [0];
for (Account acc : accs)
{
cons.add (new Contact (lastName=acc.name + '1', accountid=acc.id));
cons.add (new Contact (lastName=acc.name + '2', accountid=acc.id));
}
insert cons;
SOQL
Query
Variable
Declaration
Control
Structure
Array
Data
Operation
Commit
Transaction

41.
Apex code is stored as metadata, interpreted at runtime, and
cached for scalability

42.
§ Bulk DML
§ Email and messaging
§ Asynchronous processing (@future)
§ XmlStream / HTTP (RESTful) services classes
§ Declarative exposure as new Web Services
§ Platform Encryption
Apex is deeply integrated with platform features

43.
Share Your Feedback, and Win a GoPro!
3 Earn a GoPro prize entry for
each completed survey
Tap the bell to take a
survey2Enroll in a session1