Microservices - SOA reminded of what it was supposed to deliver?

Microservices - SOA reminded of what it was supposed to deliver?

1. 1. Microservices SOA reminded of what it was supposed to deliver? Jeppe Cramon - @jeppec μService Conference London 2014
2. 2. SOA or Microservices? Higher reusability, significant reduction of development cost, increased reliability, improved maintainability, enhanced quality… … building blocks that can be reused and tweaked as needed. There’s also a redundancy benefit: Should one … fail the other ones will work in the app. Better ROI, More Security, Fewer Defects, Reusable assets, Better Maintainability, More reuse, Better parallelism in development, Better scalability, Higher availability….
3. 3. For years no one talked about SOA Mostly because of the many failures… Example: I've used service-oriented architecture for the better part of a decade, and I can honestly report the following … • I have never seen the "contract" feature of any service I've written or designed be leveraged to enable its re-use • I have never seen a service from one system be recoupled to another • I have never seen encapsulated logic used in multiple applications Source: http://www.techrepublic.com/blog/tech-decision-maker/the-unfulfilled-promise-of-soa/
4. 4. In my opinion SOA is not to blame for the dissapointment
5. 5. I believe that Microservices is a welcome opportunity to brush off the dust, learn form the past and perhaps improve
6. 6. Beware of sales pitches
7. 7. The fine-grained, stateless, self-contained nature of microservices creates decoupling between different parts of a code base and is what makes them easy to update, replace, remove, or augment. Microservices: The resurgence of SOA principles and an alternative to the monolith
8. 8. Trying hard to pretend it’s not difficult to develop distributed systems
9. 9. It seems the argument is that just by making things fine grained We by magic achieve decoupling
10. 10. Decoupled? Source: https://qconsf.com/system/files/presentation-slides/QConSF-MicroServices-IPC-Netflix-Sudhir-2014.pptx
11. 11. Monoliths are often blamed for being the lasagna of software due to coupling Spaghetti in layers
12. 12. So if we keep up the same habits that resulted in messy monoliths What’s makes us think that we won’t end up with
13. 13. Microservice spaghetti?
14. 14. Most “Services” today are built on top of Monoliths DB DB Service A.4 Monolith A Monolith B Monolith C Monolith D Service A.1 Service A.2 Service A.3 Service B.1 Service B.2 Service B.3 Service B.4 Service D.1 Service D.4 Service D.3 Service D.2 Service C.4 Service C.3 Service C.2 Service C.1 DB DB
15. 15. If we create naïve microservices DB DB Service A.4 DB Monolith A Monolith B Monolith C Monolith D Service A.1 Service A.2 Service A.3 Service B.1 Service B.2 Service B.3 Service B.4 Service D.1 Service D.4 Service D.3 Service D.2 Service C.4 Service C.3 Service C.2 Service C.1 DB DB DB DB DB DB DB DB DB DB DB DB DB DB DB DB DB
16. 16. Then we’ve IMO just gone from bad
17. 17. To Worse
18. 18. So does that mean that breaking things apart is a bad thing?
19. 19. Definitely NOT! But we need some guidelines to avoid things becoming messy (once again)
20. 20. So how do we get there? Let’s look at some of the characteristics of Microservices (that most seem to agree upon)
21. 21. Microservice characteristics • Single Responsibility Principle (SRP) • Small • Own process • Valuable • Replaceable • Upgradeable • Independent • Encapsulated • Composable • Testable • Fast startup/shutdown • Client friendly
22. 22. Single Responsibility Principle (SRP) Billing DB Order Fulfillment DB SMS Gateway Shipping DB Management Reporting DB A Service should have a single responsibility and this responsibility should be entirely encapsulated within the service
23. 23. Does a Microservice own its data? • Remember layered SOA? • IMO a service that only deals with data persistence is best called a database/datastore • It already has a nice API – We don’t need to bubble wrap it with REST or Async messages • Don’t split the atom – we need cohesion as well as decoupling! • If we want datastore abstraction (so we can swap out Postgresql with Mongo or Redis) there this little pattern called Respository.
24. 24. So how big/small should a microservice/responsibility be?
25. 25. But I’ve heard that a “Microservice should be no larger than 100 Lines of Code!?”
26. 26. Enough already! When did we get away from solving business problems and start talking technology, JSON over XML, REST, LoC’s, Frameworks again?
27. 27. It reminds me of the early SOA days All that was talked about then was which ESB to purchase, to how set it up, etc. No body discussed the problems that the ESB was supposed to solve (or when it was appropriate) or what services to build. We were caught in technology silver bullet again!
28. 28. We kind of forgot that SOA was about business agility It was too much “Build it and they will come”
29. 29. There is value in making things smaller For one thing it easier to reason about them in isolation
30. 30. Going small has its advantages Source: @aviranm
31. 31. SRP This sounds really good but what about cross Service relationships? Customer Orders Products
32. 32. Beware… When we break up big things into small pieces we invariably push the complexity to their interaction. Michael Feathers https://michaelfeathers.silvrback.com/microservices-until-macro-complexity
33. 33. Should we replace joins with RPC?
34. 34. Change Address Response Consumer Change Address Request Provider RPC or Request/Response - Synchronous 2 way communication Remote Procedure Call Change Address Request Change Address Request VS. Request Channel Consumer Provider Change Address Reply Change Address Reply Reply Channel Request/Reply – Asynchronous 2 way communication
35. 35. Size this and Size that! Be Careful If Microservices are good, then Nanoservices must be even better? Why not one-liner services?
36. 36. Nano Services Unless we have a very reason for doing so, we risk building services that are so fine-grained that their costs outweigh their utility* *Read Arnon Rotem-Gal-Oz’s Nano Services Anti Pattern:http://arnon.me/wp-content/uploads/2010/10/Nanoservices.pdf
37. 37. Microservices are valuable The value of a microservice must exceed the cost of building & operating it. Microservices entail costs for serializations, deserializations, security, communication, maintenance, configuration, deployment, monitoring, etc.
38. 38. Too small services With too small services there’s a big risk that what’s left of coherence goes out the window and we get: • Communication-related and Layered coupling – E.g. Logic and persistence are not the same service • Temporal coupling – Our service can not operate if it is unable to communicate with the services it depends upon • Behavioral coupling – Our services assume behaviors with regards to what and who. – The sender/client determines what to do and knows something about how the receiver/server should satisfy the request
39. 39. Microservices == distributed objects? Service star chart
40. 40. Service Dependencies Source: https://qconsf.com/system/files/presentation-slides/QConSF-MicroServices-IPC-Netflix-Sudhir-2014.pptx
41. 41. All these services needs to be integrated And often in the form of Request/Response calls (and often over the network – aka RPC)
42. 42. Synchronous Remote Calls can lower our Fault tolerance • When servers crashes • When databases are down • When deadlocks occurs in the database – Do you retry? With synchronous RPC style Services interaction we can easily loose business data unless we use an Orchestration engine or find another way to achieve call/retry semantics, combined with compensations and lifecycle handling. Note: Automatic retry of calls only works if our Service operations are idempotent.
43. 43. Also remember: REST isn’t magic!
44. 44. What’s less fat – XML or JSON? • They’re equally slim and fast! • http://balisage.net/Proceedings/vol10/html/ Lee01/BalisageVol10-Lee01.html
45. 45. RPC and Distributed computing • As soon as a service request/response calls to another service across the network we must adhere to the laws of distributed computing*. – Reliability, Latency, Bandwidth, Transportation costs, Security • If the other services is not available then my service cannot conduct its business, which minimizes my services autonomy * See http://www.rgoarchitects.com/Files/fallacies.pdf for a walkthrough of the 8 fallacies of distributed computing
46. 46. Service autonomy Service B Service C Service A System X Service Service A System Y B Service C System X Slow/unreliable network Different SLA Slow system
47. 47. A distributed system is one where a machine I’ve never heard of can cause my program to fail. — Leslie Lamport
48. 48. Availability goes down (without additional instances of each service) Service A Service B Service C Availability: 99% Availability: 99% Availability: 99% Combined availability: 97%
49. 49. Decide if you can live with the consequences of coupling services to each other using Request/Response Different situations – different tradeoffs
50. 50. SOA was about business agility through IT/Business alignment
51. 51. Which require maturity • Organizational maturity
52. 52. Be aware of Conways Law “organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations” Teams are typically aligned with Systems/Projects and NOT with Services as I believe they should
53. 53. Which require maturity • Organizational maturity • Technical/operations maturity* – DevOps – Build – Test – Deployment – Monitoring – Etc. * See: http://highscalability.com/blog/2014/4/8/microservices-not-a-free-lunch.html
54. 54. We need a strong foundation for development to achieve this Less aligned Alignment Highly aligned 11% ”Alignment Trap” +13 74% 7% ”IT enabled Growth” -6 8% -14 +35 ”Maintenance Zone” ”Well-Oiled IT” -2 -15 +11 Less effective Effciency Highly effective % of the 504 respondents % difference compared to the overall averages IT spending Source: Bain Analysis Combined yearly growth-rate over a 3 year period +0
55. 55. To break things up We need to understand our business and its processes - How is data flowing?
56. 56. In order to build the right boundaries
57. 57. Claim Most of us are trained to think in Entities and Structural models and this is where we get our selves into trouble. We create big entangled models and forget about model boundaries and different consistency level requirements!
58. 58. The mental capacity required to understand big domain models is huge
59. 59. Many perspectives on data Online Retail System Product Unit Price Promotional Price Promotion End Date Stock Keeping Unit (SKU) Quantity On Hand (QOH) Location Code Price Quantity Ordered Name The lifecycle of the data is VERY important! Customer Pricing Inventory Sales Management Reporting
60. 60. Different perspectives on entities With in a given Domain, e.g. Retail, there will exist multiple bounded contexts/sub-domains/business capabilities such as: – Product management – Purchase – Sales – Pricing – Inventory – Shipping – Support – Accounting – Management Each of these lines of business have very specific and unique needs which are relevant for them alone in order to conduct their business. They might use the same name for the entities they’re interested in or they might use different names for the same logical entity.
61. 61. Retail domain split into a Macro architecture Billing Bounded Context Product Catalogue Bounded Context Shipping Bounded Context Sales Bounded Context Inventory Bounded Context Pricing Bounded Context
62. 62. What’s a macro architecture • It’s the static/stable(r) parts of your architecture – Which are very costly to refactor and change • Closely aligns business capabilities / bounded contexts with services
63. 63. DDD doesn’t enforce problem domain and solution domain alignment for Bounded Contexts Which is why I prefer to use the term Business Capability
64. 64. Service and Business Capability alignment “The advantage of business capabilities is their remarkable level of stability. If we take a typical insurance organisation, it will likely have sales, marketing, policy administration, claims management, risk assessment, billing, payments, customer service, human resource management, rate management, document management, channel management, commissions management, compliance, IT support and human task management capabilities. In fact, any insurance organisation will very likely have many of these capabilities.” See http://bill-poole.blogspot.dk/2008/07/business-capabilities.html
65. 65. Don’t split into services too early Cross boundary refactoring is costly
66. 66. A Service is • The technical authority for a given business capability • It is the owner of all the data and business rules that support this business capability – everywhere • It forms a single source of truth for that capability • This form of business and IT alignment ensures that we can maintain service Autonomy & Encapsulation
67. 67. So where does that leave microservices?
68. 68. Service and deployment A service needs to be deployed everywhere its data is needed • A Service represents a logical boundary • Logical responsibility and physical deployment of a service DOES NOT have to be 1-to-1 – It’s too constraining – We need more degrees of freedom – Philippe Krutchen 4+1 views of architecture: Logical and Physical designs should be independent of each other
69. 69. Service 1..* A Service is the technical authority of a specific Business Capability e.g. Sales, Shipping, Billing Service Microservice Is implemented by Services are the implementation of business processes. Business processes can span multiple services, but there will always be a single service that is the actual authority on the business process.
70. 70. Service Also known as Autonomous Components 1..* Service Microservice Is implemented by Microservices are a division of Services along Transactional boundaries (a transaction stays within the boundary of a Microservice) Microservices are the individually logical deployable units of a Service with their own Endpoints. Could e.g. be the split between Read and Write models (CQRS) - each would be their own Microservice
71. 71. Microservices are logical deployable units That doesn’t mean they HAVE to be deployed individually. Design for Distribution But take advantage of locality
72. 72. Services are the corner stone • We talk in terms of services, business capabilities and the processes/usecases they support • Microservices are an implementation detail • Microservices are much less stable (which is a good thing – it means they’re easier to replace) • A microservice in one service can subscribe to events from another Service or call operations on another Service (which again is implemented/supported by an internal microservice)
73. 73. Coupling matrix* Emergency services Distributed 3 layer Event oriented Command oriented Behavioral coupling High Temporal coupling Low High Low * Modified version of Ian Robinson’s matrix: http://iansrobinson.com/2009/04/27/temporal-and-behavioural-coupling/
74. 74. Using Business Events to drive Business Processes Online Ordering System Sales Service Billing Shipping Customers Sales Bus Web Shop (Composite UI) Billing Service Shipping Service Order Accepted Event AcceptOrder Command The sales fulfillment processing can now begin…
75. 75. Cascading events give rise to business processes
76. 76. Event driven process Sales Service Online Ordering System Order Accepted Billing Service Orderfulfilment Service Process Manager/ Saga/ Orchestration-Engine Shipping Service Bus Order Accepted Customer Billed Order Accepted Customer Billed Order Authorize d Order Authorize d Works as a Finite State Machine (WorkFlow) handling the life cycle of Shipping and thereby forms a very central new Aggregate in the System
77. 77. This form of architecture is called an Event Driven Architecture (EDA)
78. 78. SOA and EDA are two sides of the same coin
79. 79. So are Microservices the future? Gartners Pace layered Application strategy: • Systems of Record — Established packaged applications or legacy home-grown systems that support core transaction processing and manage the organization's critical master data. The rate of change is low, because the processes are well-established, common to most organizations, and often are subject to regulatory requirements. • Systems of Differentiation — Applications that enable unique company processes or industry-specific capabilities. They have a medium lifecycle (one to three years), but need to be reconfigured frequently to accommodate changing business practices or customer requirements. • Systems of Innovation — New applications that are built on an ad hoc basis to address new business requirements or opportunities. These are typically short lifecycle projects (zero to 12 months) using departmental or outside resources and consumer-grade technologies. • "These layers correspond to the notion of business leaders having common ideas, different ideas, and new ideas,"
80. 80. Conclusions • In my opinion Microservices is one SOA delivery model • Focus should be on aligning Services to Business Capabilities – Make service as small as possible and as big as necessary • Beware of the coupling matrix – avoid synchronous communication between services unless you’re ready to pay the price • Implement Services as a set of Microservices that are logically deployable and divide them along transactional boundaries • Be aware of the requirements for organizational, business and technical maturity • If we can get there then I believe Microservices will be able to help us achieve a some of what SOA promised, but due to different factors rarely delivered
81. 81. Thank you  Jeppe Cramon TigerTeam www.tigerteam.dk www.tigerteam.dk/category/soa/ www.tigerteam.dk/category/soa/microservices/ Twitter: @tigerteamdk and @jeppec