Enterprise architectures never sleep because cloud-first strategies must also become multi-cloud-first strategies. Public cloud providers such as Microsoft Azure are providing compelling services and pricing. And, most enterprises now consider their own datacenter a private cloud.
This is not a one-cloud playing field and enterprise architects must develop strategies, standards, and policies about how their data is being used, moved, and created across multiple cloud infrastructures.
Join Pivotal’s Jag Mirani and Mike Stolz along with guest, Forrester Vice President and Principal Analyst, Mike Gualtieri, as they examine the trends driving multi-cloud adoption and more importantly how to architect technical solutions to make data free to roam among them safely.
Speakers:
Mike Gualtieri, VP, PRINCIPAL ANALYST, Forrester
Jag Mirani, Product Marketing, Data Services, Pivotal
Mike Stolz, Product Lead, GemFire, Pivotal
2. Cover w/ Image
Topics
■ What’s driving multi-cloud?
■ What is the data challenge?
■ How does design thinking change in a
multi-cloud architecture?
■ What are the architectural/imperatives for
multi-cloud?
■ What are some real-world multi-cloud
use cases?
■ Key Pivotal solution components
4. Business Drivers for Multi-cloud
● Avoid vendor lock-in
● Meet quality of service requirements (online availability and response time) using multiple
distributed data centers for geographic proximity to customers and consumers
● Organizational boundaries (ex: align the tech stack and IT operations by business unit)
● Risk diversification / mitigation
● Data sovereignty, laws, regulations
● Leverage cloud provider strengths and innovation
6. What are the Data Challenges?
Overcoming Data Gravity in Multi-Cloud Enterprise Architectures
7. Data Gravity in the Enterprise
● New data is being generated in the cloud, outside the walls of the enterprise
● Data sources are becoming more diverse
● Network bandwidth and latency
● Volume of data is still exploding
● Data distribution vs. consistency
● Data governance, laws, security, provenance
● Metadata creation and accumulation
● Failure states of the system
And it’s not all internet data ...
14. How does design thinking change in a multi-
cloud architecture?
Overcoming Data Gravity in Multi-Cloud Enterprise Architectures
15. ● Weigh the cost-benefit of multicloud portability for each application;
prioritize accordingly. Segment applications based on primary need:
redundancy vs. functional distribution.
● Avoid the factors that contribute to lock-in
● Design for cloud native environments, favoring modular design with
contextual isolation and statelessness (12-factor apps)
● Map the workload requirements for each application (or components of
each application) to the cloud provider that provides the best-of-breed
services
● Assess the culture and appetite for formalizing a multi-cloud strategy
Application Design Thinking for Multi-Cloud
16. What are some real-world multi-cloud use
cases?
Overcoming Data Gravity in Multi-Cloud Enterprise Architectures
17. Common Use Cases for Multi-cloud
1. Disaster Recovery
2. Public cloud as an
extension of the
datacenter
3. Active/Active
WAN Replication across
Foundations, across
Clouds
18. Disaster Recovery (DR) Restoration Pattern
● Recovery site brought online as needed
● Multiple foundations can share a recovery site
● Recovery site can reside on-premises, in a co-location
facility, or the public cloud
● Recovery site includes an operational foundation, with only
the most critical apps
● Primary site’s data is replicated to recovery site via Pivotal
Cloud Cache’s WAN replication
● Can be used in conjunction with other methods
19. Public Cloud as Extension of the Datacenter
● For short periods of time to offload spikes in traffic
● Often in support of major business events
(product launch, marketing campaign, or surge in
seasonal traffic)
● Pay for extra resources only when they are
needed
● Requires a high-speed, dedicated connection
● WAN replication propagates data changes in both
directions
20. Active - Active Deployment
● Global traffic manager directs traffic from clients
● Users can be routed to the PCF foundation physically closest to them
● Other routing policies: round-robin, weight-based, latency-based, geolocation, and session affinity
(cookie-based or client IP)
● PCC Wan replication propagates data changes in both directions
WAN
Replication
21. What are the Architectural Imperatives for
Multi-Cloud?
Overcoming Data Gravity in Multi-Cloud Enterprise Architectures
23. Conflict Resolution in Active/Active Setup
23
● PCC automatically detects conflicts and retains the latest
data
○ Local timestamps and conflict detection algorithms
● Can use custom code for conflict resolution
● Alternative: design the system to avoid conflicts ...
25. Multi-site Active-Active Design Patterns
1. Exchange Pattern
NYSE
LSE
LSE
TSE
NYSE, TSE Read--only
LSE, TSE Read--only
NYSE, LSE Read--only
Client connects
to all
exchanges it
needs for
writing, uses
local copy for
read only
access.
26. Multi-site Active-Active Design Patterns
2. The "Realm Manager"
Pattern:
Use the “Command”
pattern to request that
an action be performed
on your behalf.
Request gets forwarded
to all distributed
systems but only the
one with the right
permission actually
takes the action.
Read Only For This Customer
Read Only For This Customer
Write Permission For This Customer
27. Multi-Site Active-Active Design Patterns
3. Follow the Sun
Pattern:
This is the "Global book"
pattern common in
Financial Services.
The token is here
28. Multi-Site Active-Active Design Patterns
4. Inventory Allocation Pattern:
This pattern is
commonly used when
there are multiple
trading venues and
selling short is not
allowed.
Partial Inventory
Partial Inventory
Partial Inventory
Partial Inventory
29. Multi-Site Active-Active Design Patterns
5. Apology based computing:
This is the pattern
that Max Feingold
refers to when he
says:
“At global scale,
getting the truth is
really really
expensive.”
30. Key Pivotal Solution Components
Overcoming Data Gravity in Multi-Cloud Enterprise Architectures
32. Pivotal Cloud
Cache
● Cross DC data sharing
● Dev can push server-side code to save data to backing store
● Support persistence w/Regions
● Support multi-WAN connected cluster
33. Pivotal Cloud Foundry Marketplace
• Easy accessibility
through Marketplace
• Instant Provisioning
• Bind to apps through
easy to use interface
• Lifecycle management
• Common access
control and audit trails
across services
MySQL New Relic
Single Sign-
On
RabbitMQ
Config
Server
Service
Directory
Circuit
Breaker
Signal
Sciences
Crunchy
PostgreSQL AND
MORE
Services Marketplace
Pivotal Cloud
Cache
Dynatrace
Extending the Pivotal Cloud Foundry Platform for Microservices Architectures
34. Multi-Cloud is Inevitable
● Enables flexibility and choice
○ Go in with a well considered multicloud strategy and
plan, rather than ad-hoc
● Map cost-benefit back to business drivers: business
continuity, portability and the absence of lock-in,
opportunistic use case placement and future-proofing, ...
35. Summary: Assessing Your Choices
● Option 1: Build directly on top of an IaaS
○ Prepare (cross train) staff on all identified cloud providers
○ Choose native management tools and operational processes for each
cloud
○ Maintain diligence towards avoiding lock-in
● Option 2: Build on top of a PaaS like Pivotal Cloud Foundry
○ Platform, tools, and methodology that mask the differences between IaaS
○ Continuous and rapid provisioning of apps and services
○ Automated ‘day 2’ operations