A talk on Container networking, and Distributed Systems in Erlang

1. © 2016 Mesosphere, Inc. All Rights Reserved.
ERLANG &
CONTAINER
NETWORKING
@SARGUN
1
Sargun Dhillon, August 2016

2. © 2016 Mesosphere, Inc. All Rights Reserved.
WHO AM I?
2

3. © 2016 Mesosphere, Inc. All Rights Reserved.
AGENDA
3
• A brief history of how we got here
• How we’re trying to make it better
• Some systems we’ve built to enable us to better ship
• Why Erlang helped us get there

4. © 2016 Mesosphere, Inc. All Rights Reserved.
A BRIEF
HISTORY OF
NETWORKS
IN THE DC
4

5. © 2016 Mesosphere, Inc. All Rights Reserved.
ORGANIZATIONS CIRCA 2007
5
•Before DevOps was first heard
•Clear differentiation of ownership
•The datacenter was owned by a the NOC
•Deployment of services was done by sysadmins in the operations group
•Developers operated without access to production
•Production deployments gated by QA, Operations

6. © 2016 Mesosphere, Inc. All Rights Reserved.
SOFTWARE CIRCA 2007
6
•Different services glued together via CORBA, XML-RPC, SOAP
•No one was really consciously doing microservices
•Networks were static, giant layer 2 domains
•Load Balancing provided by hardware
•Firewall provided by hardware
•Everyone ran their own datacenter
•EC2 in its infancy, only a year prior has the term “Cloud” began to become
popular
•Systems statically partitioned

7. © 2016 Mesosphere, Inc. All Rights Reserved. 7
SaaS continued to grow at
an incredible rate

8. © 2016 Mesosphere, Inc. All Rights Reserved. 8
There became a race to ship faster

9. © 2016 Mesosphere, Inc. All Rights Reserved. 9
We kept the software alive
By feeding it
With Sysadmins

10. © 2016 Mesosphere, Inc. All Rights Reserved. 10
We kept the machines alive
By feeding them
With Blood

11. © 2016 Mesosphere, Inc. All Rights Reserved. 11
This wasn’t working

12. © 2016 Mesosphere, Inc. All Rights Reserved.
ORGANIZATIONS 2008+
12
•We began seeing a gradual shift in the industry where lines between QA, Dev, and
Ops were blurring
•Devops term coined in 2008, first DevOpsDay in 2009
•Gradual adoption of the cloud, fewer organizations owning their own datacenters
•Either networking was outsourced to the cloud, or typically remained in a small
internal organization
•Needed to reduce ratio of operators to servers

13. © 2016 Mesosphere, Inc. All Rights Reserved.
SOFTWARE CIRCA 2008+
13
•Popularization of Open Source tooling to automate much of traditional operations, and QA
•Jenkins / Hudson
•Puppet / Chef
•Capistrano
•Popularization of stacks requiring with more complex operational requirements
•Nutch / Hadoop
•NoSQLs
•Still statically partitioned machines
•Networks still sacred territory

14. © 2016 Mesosphere, Inc. All Rights Reserved.
CIRCA 2011
14
•Much of what’s been happening for the past half-decade hits networking
•Much of this falls under the term “SDN” (Software Defined Networking) or “NFV” (Network
Function Virtualization)
•Hastened by the adoption of VMs in the enterprise in the hype cycle
•Openflow promises to fix everything
•Major adoptions of the cloud by startups as well as enterprise
•Virtualization begins to become mainstream as a mechanism of consolidating workloads
•The invention of the “private cloud”
•DotCloud / Docker funded by Y-combinator a year earlier
•Term “Microservice” coined

15. © 2016 Mesosphere, Inc. All Rights Reserved.
CIRCA 2013
15
• Docker becomes instant hit and brings containers to the forefront
• Dynamic partitioning begins to make in-roads
• Google releases Omega paper
• Apache Aurora open sourced
• Microservice counts explode, demanding collocation of workloads for efficiency
• Mesosphere Founded
• Site Reliability Engineering begins to popularize and further blur the lines between
Dev, Ops, and QA

16. © 2016 Mesosphere, Inc. All Rights Reserved. 16
Everything was changing

17. © 2016 Mesosphere, Inc. All Rights Reserved. 17
Why?

18. © 2016 Mesosphere, Inc. All Rights Reserved. 18
Business Value

19. © 2016 Mesosphere, Inc. All Rights Reserved. 19

20. © 2016 Mesosphere, Inc. All Rights Reserved.
BENEFITS
20
•Reduction in cost of goods sold
•Smaller engineer to server ratio
•Linear, or super linear growth rate of engineering team to servers is unsustainable
•Smaller engineer to capability ratio, where capability includes:
•Features
•Throughput
•Better User Experience
•Better availability
•Quicker release to features

21. © 2016 Mesosphere, Inc. All Rights Reserved. 21
But at what cost?

22. © 2016 Mesosphere, Inc. All Rights Reserved. 22
Complexity

23. © 2016 Mesosphere, Inc. All Rights Reserved.
OLD WORLD
23

24. © 2016 Mesosphere, Inc. All Rights Reserved.
NEW WORLD
24

25. © 2016 Mesosphere, Inc. All Rights Reserved.
DIVING DEEPER
25

26. © 2016 Mesosphere, Inc. All Rights Reserved. 26
It was time for something completely
new

27. © 2016 Mesosphere, Inc. All Rights Reserved.
ENTER
DC/OS
27

28. © 2016 Mesosphere, Inc. All Rights Reserved.
CONTAINERS
WITH
BATTERIES
INCLUDED
28
• Includes everything you need to move away from
2007
• Networking
• Security
• Application deployment, and orchestration
• Stateful services

29. © 2016 Mesosphere, Inc. All Rights Reserved. 29
In Your Datacenter

30. © 2016 Mesosphere, Inc. All Rights Reserved. 30
As a Black Box

31. © 2016 Mesosphere, Inc. All Rights Reserved. 31
But, our story starts:
~November 2015
(~8 months ago)

32. © 2016 Mesosphere, Inc. All Rights Reserved. 32
Networking Features

33. © 2016 Mesosphere, Inc. All Rights Reserved. 33
Service Discovery
(Mesos-DNS)

34. © 2016 Mesosphere, Inc. All Rights Reserved. 34
…Ish

35. © 2016 Mesosphere, Inc. All Rights Reserved. 35
DNS SRV Based Service Discovery

36. © 2016 Mesosphere, Inc. All Rights Reserved. 36
Everyone has DNS right?

37. © 2016 Mesosphere, Inc. All Rights Reserved. 37
And GLibc even has a bug open for it!

38. © 2016 Mesosphere, Inc. All Rights Reserved. 38
…Opened in 2005

39. © 2016 Mesosphere, Inc. All Rights Reserved. 39
We needed a solution that actually
worked

40. © 2016 Mesosphere, Inc. All Rights Reserved. 40
So, we performed an OODA loop
1.Observe
2. Orient
3. Decide
4. Act

41. © 2016 Mesosphere, Inc. All Rights Reserved.
OBSERVE: SERVICE DISCOVERY
41
• Existing Dynamic Service Discovery Solutions:
• Etcd
• Finagle + Zookeeper
• Consul
• Existing Static Service Discovery Solutions:
• Amazon ELB
• Hardware Load Balancers
• Service Discovery is an afterthought
Gathering our data about the field

42. © 2016 Mesosphere, Inc. All Rights Reserved.
OBSERVE: NETWORKING
42
• Everybody assumes IP per application instance
• Everybody assumes reliable DNS
• Some people want to be fast
• Some people want security
• Nobody wants to edit application code
• Nobody wants to talk to their network engineer
Gathering our data about the field

43. © 2016 Mesosphere, Inc. All Rights Reserved.
ORIENT:
DC/OS
NETWORKING
CORE TENANTS
43
•DC/OS must be agnostic to the underlying environment
•AWS / Azure / GCE / Softlayer as the lowest common
denominators
•DC/OS should require no to minimal changes to the code in
order to work
•DC/OS should provide similar services to existing
environments
•Fixed load balancers
•Security
•IP/Container
•We do not want to require a change in organization
procedures
•We want to be secure

44. © 2016 Mesosphere, Inc. All Rights Reserved.
DECIDE
44
What do we want to build?
• Load Balancing
• Seamless Service Discovery
• Reliable DNS
• External cluster Access
• Metrics and Discoverability
• IP Per Container

45. © 2016 Mesosphere, Inc. All Rights Reserved.
DC/OS ARCHITECTURE (~NOV ’15): POLYGLOT MICROSERVICES
45

46. © 2016 Mesosphere, Inc. All Rights Reserved. 46
What Did We Build?

47. © 2016 Mesosphere, Inc. All Rights Reserved.
ACT:
DEVELOPED
NEW ERLANG
SERVICES
47
• Service Discovery
• Navstar
• Spartan
• Load Balancing
• Minuteman
• Networking API
• Fishladder
• Control Plane
• Lashup

48. © 2016 Mesosphere, Inc. All Rights Reserved.
DC/OS ARCHITECTURE (TODAY): POLYGLOT MICROSERVICES
48

49. © 2016 Mesosphere, Inc. All Rights Reserved.
CODE METRICS
49
~14,000 LoC
118 Modules
7 Repos
1.5 Engineers
0.5 Consultants

50. © 2016 Mesosphere, Inc. All Rights Reserved. 50
DNS:
Spartan

51. © 2016 Mesosphere, Inc. All Rights Reserved. 51
HA DNS

52. © 2016 Mesosphere, Inc. All Rights Reserved. 52
/etc/resolv.conf
nameserver 8.8.8.8
nameserver 8.8.4.4
Normal DNS

53. © 2016 Mesosphere, Inc. All Rights Reserved. 53
Ideal DNS Request

54. © 2016 Mesosphere, Inc. All Rights Reserved. 54
Ideal DNS Request

55. © 2016 Mesosphere, Inc. All Rights Reserved. 55
But, sometimes things go wrong

56. © 2016 Mesosphere, Inc. All Rights Reserved. 56
Not so Ideal DNS Request

57. © 2016 Mesosphere, Inc. All Rights Reserved. 57
Wait at least 1 second

58. © 2016 Mesosphere, Inc. All Rights Reserved. 58
Not so Ideal DNS Request

59. © 2016 Mesosphere, Inc. All Rights Reserved. 59
Wait at least 1 second

60. © 2016 Mesosphere, Inc. All Rights Reserved. 60
Not so Ideal DNS Request

61. © 2016 Mesosphere, Inc. All Rights Reserved. 61
Wat?

62. © 2016 Mesosphere, Inc. All Rights Reserved. 62
Spartan DNS

63. © 2016 Mesosphere, Inc. All Rights Reserved. 63
Spartan DNS

64. © 2016 Mesosphere, Inc. All Rights Reserved. 64
Spartan DNS

65. © 2016 Mesosphere, Inc. All Rights Reserved. 65
Spartan Controls Timeouts

66. © 2016 Mesosphere, Inc. All Rights Reserved. 66
Edge DNS

67. © 2016 Mesosphere, Inc. All Rights Reserved. 67

68. © 2016 Mesosphere, Inc. All Rights Reserved.
SPARTAN:
BENEFITS
68
•Lowers Average Latency
•Can control timeouts to sub-second
•Many DNS queries don’t need to go off-node

69. © 2016 Mesosphere, Inc. All Rights Reserved. 69
Load Balancing:
Minuteman

70. © 2016 Mesosphere, Inc. All Rights Reserved. 70
The Players

71. © 2016 Mesosphere, Inc. All Rights Reserved. 71
Setting it up

72. © 2016 Mesosphere, Inc. All Rights Reserved. 72
Intercepting the
connection

73. © 2016 Mesosphere, Inc. All Rights Reserved. 73
Remap

74. © 2016 Mesosphere, Inc. All Rights Reserved. 74
Success

75. © 2016 Mesosphere, Inc. All Rights Reserved.
MINUTEMAN:
BENEFITS
75
•Appearance of a fixed-load balancer
•Fully distributed
•Other than first packet, the entire lifetime is handled
in kernel space

76. © 2016 Mesosphere, Inc. All Rights Reserved. 76
How do we tie it all together?

77. © 2016 Mesosphere, Inc. All Rights Reserved.
GLOBAL STATE
77
• Load Balancer Task Mapping
• DNS Zones
• Virtual Network Routing Tables
• Reachability
• Security ACLs

78. © 2016 Mesosphere, Inc. All Rights Reserved. 78
Computers

79. © 2016 Mesosphere, Inc. All Rights Reserved. 79
Constant Churn

80. © 2016 Mesosphere, Inc. All Rights Reserved. 80
Sometimes
They Break

81. © 2016 Mesosphere, Inc. All Rights Reserved. 81
Sometimes
Many Break

82. © 2016 Mesosphere, Inc. All Rights Reserved. 82
Sometimes
You don’t know

83. © 2016 Mesosphere, Inc. All Rights Reserved.
HOW DO YOU
DO
SIGNALING?
83

84. © 2016 Mesosphere, Inc. All Rights Reserved. 84
Nobody ever got fired for using
Zookeeper

85. © 2016 Mesosphere, Inc. All Rights Reserved. 85
We know about Zookeeper

86. © 2016 Mesosphere, Inc. All Rights Reserved. 86
It works…usually

87. © 2016 Mesosphere, Inc. All Rights Reserved. 87
How else can we do this?

88. © 2016 Mesosphere, Inc. All Rights Reserved. 88
Naively
(See: Distributed Erlang)

89. © 2016 Mesosphere, Inc. All Rights Reserved. 89
Massive Amount of Information

90. © 2016 Mesosphere, Inc. All Rights Reserved. 90
Who else has solved this?

91. © 2016 Mesosphere, Inc. All Rights Reserved. 91
Academia

92. © 2016 Mesosphere, Inc. All Rights Reserved. 92
Control Plane:
Lashup

93. © 2016 Mesosphere, Inc. All Rights Reserved. 93
How do we scale the naive approach?

94. © 2016 Mesosphere, Inc. All Rights Reserved. 94
All we need is a sparse, connected graph

95. © 2016 Mesosphere, Inc. All Rights Reserved. 95
But how?

96. © 2016 Mesosphere, Inc. All Rights Reserved. 96
Enter: HyParView
Builds a connected graph (overlay), where the degree is <=K

97. © 2016 Mesosphere, Inc. All Rights Reserved. 97
Boot Time

98. © 2016 Mesosphere, Inc. All Rights Reserved. 98

99. © 2016 Mesosphere, Inc. All Rights Reserved. 99

100. © 2016 Mesosphere, Inc. All Rights Reserved. 100

101. © 2016 Mesosphere, Inc. All Rights Reserved. 101

102. © 2016 Mesosphere, Inc. All Rights Reserved. 102

103. © 2016 Mesosphere, Inc. All Rights Reserved. 103

104. © 2016 Mesosphere, Inc. All Rights Reserved. 104

105. © 2016 Mesosphere, Inc. All Rights Reserved. 105

106. © 2016 Mesosphere, Inc. All Rights Reserved. 106

107. © 2016 Mesosphere, Inc. All Rights Reserved. 107

108. © 2016 Mesosphere, Inc. All Rights Reserved. 108

109. © 2016 Mesosphere, Inc. All Rights Reserved. 109

110. © 2016 Mesosphere, Inc. All Rights Reserved. 110

111. © 2016 Mesosphere, Inc. All Rights Reserved. 111
Connected Graph
Hyparview

112. © 2016 Mesosphere, Inc. All Rights Reserved. 112
Constant Adaptive
Health Checks*
*Borrowed from SWIM, Gossip Style Failure Detector

113. © 2016 Mesosphere, Inc. All Rights Reserved. 113
Dealing with Failure
Hyparview

114. © 2016 Mesosphere, Inc. All Rights Reserved. 114
Dealing with Failure
Hyparview

115. © 2016 Mesosphere, Inc. All Rights Reserved. 115
Dealing with Failure
Hyparview

116. © 2016 Mesosphere, Inc. All Rights Reserved. 116
Dealing with Failure
Hyparview

117. © 2016 Mesosphere, Inc. All Rights Reserved. 117
Routing*
*Borrowed from Dijkstra, OSPF, IS-IS

118. © 2016 Mesosphere, Inc. All Rights Reserved. 118

119. © 2016 Mesosphere, Inc. All Rights Reserved. 119

120. © 2016 Mesosphere, Inc. All Rights Reserved. 120
Just Run Dijkstra*
*BFS / DFS

121. © 2016 Mesosphere, Inc. All Rights Reserved. 121
Minimum Spanning Tree, From Sender A

122. © 2016 Mesosphere, Inc. All Rights Reserved. 122
A Wild Pokemon Appears!*
*CRDT papers
Database

123. © 2016 Mesosphere, Inc. All Rights Reserved. 123
CRDTS: Semilattices
“Time”

124. © 2016 Mesosphere, Inc. All Rights Reserved. 124
CRDT KV Store

125. © 2016 Mesosphere, Inc. All Rights Reserved. 125
“Demo”

126. © 2016 Mesosphere, Inc. All Rights Reserved. 126
“Demo”
(foo@3c075477e55e)10> nodes().
[baz@3c075477e55e,bar@3c075477e55e]
(bar@3c075477e55e)8> nodes().
[foo@3c075477e55e,baz@3c075477e55e]
(baz@3c075477e55e)7> nodes().
[foo@3c075477e55e,bar@3c075477e55e]

127. © 2016 Mesosphere, Inc. All Rights Reserved. 127
“Demo”
(foo@3c075477e55e)9> lashup_kv:value([erlang_users]).
[]
(bar@3c075477e55e)9> lashup_kv:value([erlang_users]).
[]
(baz@3c075477e55e)9> lashup_kv:value([erlang_users]).
[]

128. © 2016 Mesosphere, Inc. All Rights Reserved. 128
“Demo”
(baz@3c075477e55e)10>
net_kernel:allow([foo@3c075477e55e]).
ok
(baz@3c075477e55e)10>
net_kernel:disconnect('bar@3c075477e55e').
ok.

129. © 2016 Mesosphere, Inc. All Rights Reserved. 129
“Demo”

130. © 2016 Mesosphere, Inc. All Rights Reserved. 130
“Demo”
(bar@3c075477e55e)11>
lashup_kv:request_op([erlang_users], {update, [{update,
{number, riak_dt_pncounter}, {increment, 5}}]}).
{ok,[{{number,riak_dt_pncounter},5}]}

131. © 2016 Mesosphere, Inc. All Rights Reserved. 131
“Demo”
(foo@3c075477e55e)12> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},5}]
(bar@3c075477e55e)13> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},5}]
(baz@3c075477e55e)15> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},5}]

132. © 2016 Mesosphere, Inc. All Rights Reserved. 132
“Demo”
(foo@3c075477e55e)23> erlang:set_cookie(baz@3c075477e55e,
fake).
true
(foo@3c075477e55e)24> erlang:set_cookie(bar@3c075477e55e,
fake).
true

133. © 2016 Mesosphere, Inc. All Rights Reserved. 133
“Demo”

134. © 2016 Mesosphere, Inc. All Rights Reserved. 134
“Demo”
(foo@3c075477e55e)29> lashup_kv:request_op([erlang_users],
{update, [{update, {number, riak_dt_pncounter},
{increment, 3}}]}).
{ok,[{{number,riak_dt_pncounter},8}]}
(bar@3c075477e55e)17> lashup_kv:request_op([erlang_users],
{update, [{update, {number, riak_dt_pncounter},
{increment, 7}}]}).
{ok,[{{number,riak_dt_pncounter},12}]}
(baz@3c075477e55e)19> lashup_kv:request_op([erlang_users],
{update, [{update, {number, riak_dt_pncounter},
{increment, 11}}]}).
{ok,[{{number,riak_dt_pncounter},16}]}

135. © 2016 Mesosphere, Inc. All Rights Reserved. 135
“Demo”

136. © 2016 Mesosphere, Inc. All Rights Reserved. 136
“Demo”
(foo@3c075477e55e)1> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},26}]
(bar@3c075477e55e)25> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},26}]
(baz@3c075477e55e)23> lashup_kv:value([erlang_users]).
[{{number,riak_dt_pncounter},26}]

137. © 2016 Mesosphere, Inc. All Rights Reserved.
LASHUP KV
137
•Datatypes:
•Maps
•Composable
•Sets
•Flags
•Counters
•Registers
•Last-write-wins

138. © 2016 Mesosphere, Inc. All Rights Reserved. 138
Does Lashup provide Business Value?

139. © 2016 Mesosphere, Inc. All Rights Reserved. 139
Prior to Lashup

140. © 2016 Mesosphere, Inc. All Rights Reserved. 140
After Lashup

141. © 2016 Mesosphere, Inc. All Rights Reserved.
PROJECT
LASHUP
141
•A novel distributed systems SDK that provides:
•Membership
•Multicast Delivery
•Strongly-eventually consistent data storage
•Powers:
•Minuteman VIP dissemination
•Minuteman node liveness checks
•Overlay routing
•DNS Synchronization
•Open source: github.com/dcos/lashup

142. © 2016 Mesosphere, Inc. All Rights Reserved. 142
So,
Why Erlang/OTP?

143. © 2016 Mesosphere, Inc. All Rights Reserved. 143
Why Not?
(Go)

144. © 2016 Mesosphere, Inc. All Rights Reserved. 144
The First Version
of Minuteman
Was In
Golang

145. © 2016 Mesosphere, Inc. All Rights Reserved. 145
Stand on the Shoulders of Giants

146. © 2016 Mesosphere, Inc. All Rights Reserved. 146
What’s a good language with a healthy
set of distributed systems, and
networking libraries?

147. © 2016 Mesosphere, Inc. All Rights Reserved. 147
What’s a good language with a healthy
set of distributed systems, and
networking libraries?
With an ecosystem

148. © 2016 Mesosphere, Inc. All Rights Reserved. 148
Erlang

149. © 2016 Mesosphere, Inc. All Rights Reserved.
LASHUP KV
NEEDED A
DURABLE STORE
ANSWER:
MNESIA
149
•ACID, Transactional
•20+ years old
•Battle Tested
•Comes built-in
•Software Transaction Memory

150. © 2016 Mesosphere, Inc. All Rights Reserved.
LASHUP KV
NEEDED A
CRDT LIBRARY
ANSWER:
RIAK_DT
150
•Riak’s CRDT Library
•Used in production by others
•Battle Tested
•Property tested

151. © 2016 Mesosphere, Inc. All Rights Reserved.
LASHUP NEEDED
SECURE
DISTRIBUTION
ANSWER:
DISTERL
151
•SSL with mutual authentication is a matter of
configuration
•Used in production by others
•Battle Tested
•Comes Built-in

152. © 2016 Mesosphere, Inc. All Rights Reserved.
SPARTAN
NEEDED
A DNS SERVER
ANSWER:
ERL-DNS
152
•Runs DNSimple
•Used in production
•Battle Tested
•Some Statistics from DNSimple
•Serves 20K+ domains
•Does 20k queries/sec+

153. © 2016 Mesosphere, Inc. All Rights Reserved.
MINUTEMAN
NEEDED
A KERNEL API
LIBRARY
ANSWER:
GEN_NETLINK
153
•Liberal license
•Deployed by TravelPing
•Used in production
•Battle Tested

154. © 2016 Mesosphere, Inc. All Rights Reserved.
LASHUP
NEEDED A
GRAPH LIBRARY
ANSWER:
DIGRAPH
154
•Built-in Library to Erlang/OTP
•Correctness verified
•Simple API
•Downsides:
•Slow
•Heavy

155. © 2016 Mesosphere, Inc. All Rights Reserved. 155
So, we built our own.

156. © 2016 Mesosphere, Inc. All Rights Reserved. 156
On Testing…

157. © 2016 Mesosphere, Inc. All Rights Reserved. 157
We verified lashup_gm_route against
digraph

158. © 2016 Mesosphere, Inc. All Rights Reserved. 158
Property-Based Testing:
PropEr

159. © 2016 Mesosphere, Inc. All Rights Reserved. 159
Automatically Generate Command Set
CSet ={add_node(e), add_node(b),
add_edge(e,b), …}

160. © 2016 Mesosphere, Inc. All Rights Reserved. 160
Verify Equality
Digraph lashup_gm
add_node(e) add_node(e)
Check Equivalency
add_node(b) add_node(b)
Check Equivalency
add_edge(e, b) add_edge(e, b)
Check Equivalency

161. © 2016 Mesosphere, Inc. All Rights Reserved. 161
…Tens of Thousands of Times

162. © 2016 Mesosphere, Inc. All Rights Reserved. 162
Common Test Makes Integration Testing
A Breeze

163. © 2016 Mesosphere, Inc. All Rights Reserved.
COMMON TEST:
ON EACH
CHECK-IN
(LASHUP)
163
•Spins up 25 virtual nodes
•Torture tests:
•Partitions nodes randomly for 5 minutes to see if it
can generate a permanent split
•Kills nodes randomly to verify global health check
convergence
•Verifies multicast works even in poor network
conditions
•Writes divergent keys to all nodes, and reconnects
them
•~500 lines of tests, 80%+ code coverage

164. © 2016 Mesosphere, Inc. All Rights Reserved.
TAKE-AWAYS
164
•The container ecosystem is in is early stages
•Erlang is suitable for building modern, reliable
distributed systems
•These systems are tricky, so testing is key
•We have a healthy ecosystem
•There need to be alternatives to consensus

165. © 2016 Mesosphere, Inc. All Rights Reserved.
GET IN TOUCH
IF YOU’RE
INTERESTED
(P.S. WE’RE HIRING)
165

166. © 2016 Mesosphere, Inc. All Rights Reserved.
ERLANG &
CONTAINER
NETWORKING
@SARGUN
166