The Problem with Networking - Packet Pushers Future of Networking Summit - Interop Las Vegas 2016

Transcript

1. © 2015 Packet Pushers Interactive LLC Page Future of Networking Interop Las Vegas 2016 1
2. © 2015 Packet Pushers Interactive LLC Page Agenda 2 Seeing the Future Clearly Why the Network is broken Modern Design Concepts Standards Commoditisation
3. © 2015 Packet Pushers Interactive LLC Page Presentation Style • My presentation style is …. different • Going to hit a bunch of big ideas • and leave lots of pieces on the ground for you to pickup 3
4. Seeing The Future Clearly
5. © 2015 Packet Pushers Interactive LLC Page Enterprise Comes Last • Total Smartphone Sales = • Where is innovation happening ? • Where are the cool kids & hipsters ? 5
6. © 2015 Packet Pushers Interactive LLC Page 6
7. © 2015 Packet Pushers Interactive LLC Page 7 Source: http://ben-evans.com/benedictevans/2016/3/29/presentation-mobile-ate-the-world
8. © 2015 Packet Pushers Interactive LLC Page 8 Source: https://www.cbinsights.com/blog/iot-funding-startups-q1-2016/
9. © 2015 Packet Pushers Interactive LLC Page Define Enterprise IT • Definition: Technology is not my business. • My business in finance, manufacturing, health care etc • Characteristics • IT = cost centre (not a profit centre) • 5 - 10 year ownership cycles • Change resistant / averse • Buy brands instead of solutions 9
10. © 2015 Packet Pushers Interactive LLC Page You Are Not a Snowflake • Your network is 98% the same as any other • Same hardware, same software, same designs, same cable, same training, same tech supports • We all hit the same problems and fail for the same reasons (repeated failure) • Every human is different but compared to a lizard you are all exactly the same • Its sales-wallah to claim you are unique 10 Talk on OpenStack Failures : https://www.youtube.com/watch?v=r62yERtHaSg
11. © 2015 Packet Pushers Interactive LLC Page IOT and Networking • Internet of Things has two categories for now • industrial / corporate / multi-national • retail / consumer • Industrial/Corporate • Where: buildings, factories, • What: Sensors and Platforms • How: networking, money and brand • Retail / Consumer • Where: Home or in the hand • What: Watches, Smartphones, • How: Internet connectivity 11 http://blogs.cisco.com/news/cisco-announces-iot
12. © 2015 Packet Pushers Interactive LLC Page Industrial IOT Characteristics • Big Business • GE, Phillips, Johnson Controls • Big Money • Services play • stop selling, start renting/ taxing • Deliver: IOT Cloud platform that makes money every year 12 image: https://www.cbinsights.com/blog/top-startups-iiot/
13. © 2015 Packet Pushers Interactive LLC Page Network is broken But why ? 13
14. © 2015 Packet Pushers Interactive LLC Page 14 http://www.bloomberg.com/news/articles/2015-12-16/top-u-s-carriers-2016-spending-outlook-ominous-after-2015-cuts Source: DM Martins Research, using data from Company Reports and Yahoo Finance
15. © 2015 Packet Pushers Interactive LLC Page Vendor Revenue Changing 15 Cisco Financial Results in 2016
16. © 2015 Packet Pushers Interactive LLC Page 16
17. © 2015 Packet Pushers Interactive LLC Page Design Practices Thoughts on architecture 17
18. © 2015 Packet Pushers Interactive LLC Page Museums & Factories 18 Data Centre as a Factory Goods In - Goods Out. Fast, Value Added and Quality Industrial Production Goods In Services In Waste Products Services Out Compute Storage Network Operations Maintenance “Software Based” Machines Data Centre as a Museum Assets and Users come in. Nothing leaves or delivered Curated Display Room Assets In Compute Storage Network Operations Maintenance Monolithic Apps Users
19. © 2015 Packet Pushers Interactive LLC Page Why Open Source ? • I’m big on Open Source, here’s why: 19 Duplicated Effort © Greg Ferro 2014 common hardware components standard features common hardware components common interfaces standard features unique customer value duplicated effort Vendor A Vendor B common interfaces differentiation differentiation Shared Effort © Greg Ferro 2014 open source, shared base hardware components standard features Vendor A Vendor B “unique customer value” total customer value differentiation differentiation common interfaces
20. © 2015 Packet Pushers Interactive LLC Page Value of Networking 20 ‣ Network services is automation, integrated provisioning, asset management ‣ Not just firewalls or load balancing ‣ Fast automated provisioning instead of finger banging the CLI ‣ less ops, more design (pre- provisioning)
21. © 2015 Packet Pushers Interactive LLC Page Moving past connections 21 Connectivity Services Business Value © Greg Ferro 2015 $’s Network Budgets Today (Pre-SDN) Budget mostly spent on hardware and bandwidth for connectivity Connectivity Services Business Value Visibility Mobility Operability © Greg Ferro 2015 $’s Network Budgets Tomorrow (Post-SDN) Budget increasingly spent on software and services Analytics
22. Where are the Standards ?
23. © 2015 Packet Pushers Interactive LLC Page Standards Bodies • Why ? Interoperability • Reduce impact of device use • Dates back to days when computers cost real money • Once in a lifetime, as in get a loan from the bank (not funded by cash flow) • Where are we now ? 23 Same old People from Standards Bodies with 30 years of mistakes (experience) behind them Protocol Innovation Cycle How New Protocols and Standards Are Developed Salary- slaved Vendor Engineers From CTO (aka Political) Dept Indefinable & Unpredictable Lobbying Process involving hallways, email lists, who knows who, people who have time but no money, coffee in silicon valley etc etc Hopeful Graduate Student with almost zero experience Process reaches critical constipation point eruption of ‘content’ occurs at someones favourite standards body If no one says no, a working group is started Look for crusty, multi-decade participant to shout “we know this will never work” form the back of the room Work in Progress desperate attempts to gain support, input from idiots, input from competitive interests attempting to derail, and coping with arcane bureaucratic process Standard Agreed Vendors Evaluate if “Customers Times Money” is worth their development time, investing in new silicon and marketing/sales effort
24. © 2015 Packet Pushers Interactive LLC Page Standards Bodies • IEEE • Ethernet Alliance • Metro Ethernet Forum • IETF • Linux Foundation • Open Fog, Open-O, …. • Open Daylight, ONOS, …. 24
25. © 2015 Packet Pushers Interactive LLC Page Ethernet Alliance • Pre-standard standards • its like a pre-meeting meeting • Members only (not open) • Heavy on marketing • Only useful source of information on Ethernet futures 25 http://www.ethernetalliance.org
26. © 2015 Packet Pushers Interactive LLC Page IETF • Anyone can participate • Lots of people who shouldn’t participate • Long standing ownership of our futures • Has released only 96 full standards in 30 odd years. • e.g. IPv6 is a draft standard • Enterprise has zero 26
27. © 2015 Packet Pushers Interactive LLC Page ETSI / TMF • No idea. • Some/Few Telcos seem big on them • rumours that governance and finances is “questionable” 27
28. © 2015 Packet Pushers Interactive LLC Page ITU • Dancing government elephants • Really big issues, really slowly • safely ignored 28
29. My view on Open Source
30. © 2015 Packet Pushers Interactive LLC Page Open Source and Vendors 30 Duplicated Effort © Greg Ferro 2014 common hardware components standard features common hardware components common interfaces standard features unique customer value duplicated effort Vendor A Vendor B common interfaces differentiation differentiation Shared Effort © Greg Ferro 2014 open source, shared base hardware components standard features Vendor A Vendor B “unique customer value” total customer value differentiation differentiation common interfaces
31. © 2015 Packet Pushers Interactive LLC Page Case Study in Commoditisation Tesla & Electric Cars 31
32. © 2015 Packet Pushers Interactive LLC Page Legacy Cars • Progress 32 1960s 1970s 1980s
33. © 2015 Packet Pushers Interactive LLC Page Punctuated Equilibrium • Tesla Electric Car 33
34. © 2015 Packet Pushers Interactive LLC Page Electric Cars • Take out the old • Fluids - Engine Oil, Gearbox Oil, Fuel, Water • No Drive Train - Gearbox, Differential • No Combustion - no heat, no cooling, • Less Weight • Constant maintenance • In With The New • Software managed • Connected 34
35. © 2015 Packet Pushers Interactive LLC Page Electric vs Fuel • Common Components • Li-Ion Batteries • Electric Engines • Physical Design • Production • non-standard • non-trivial • but not impossible 35 Components Simplified Repeatable Simpler Unique Complex Repeatable Simple © Greg Ferro 2014
36. © 2015 Packet Pushers Interactive LLC Page Early Cars • Early automobiles had “chaffeurs” • they heated the spark plugs • more correctly, mechanics 36
37. © 2015 Packet Pushers Interactive LLC Page New Choices - 1 • History of networking • expensive to buy, expensive to maintain • limited vendors • artificial product mappings • Cisco Cat4700/6800 vs NX7000. • Cisco Cat3850 vs NX3000 • Argulably, none of this needed to change ….. 37
38. © 2015 Packet Pushers Interactive LLC Page New Choices - 2 • Merchant Silicon • Open Operating Systems • Disaggregated Software/Hardware 38 • Now we have choices
39. © 2015 Packet Pushers Interactive LLC Page Explaining The Network • We don’t understand what goes over the network • and may never • but we think we do • We will never know WHY it is on the network 39
40. © 2015 Packet Pushers Interactive LLC Page 40 Network Edge • where is the network edge in 2016 ? • server / smartphone ? • DC LAN • Hyperconverged/ Converged Systems ? • Public cloud instances ? • Virtualization / vSwitches • Campus • VDI, Wireless • WAN • SD-WAN, multi-topology
41. © 2015 Packet Pushers Interactive LLC Page Software Eating The World • Software is Hard, Hardware Is Easy • Hardware = hard to make, easy to use • Software = easy to make, hard to use 41
42. © 2015 Packet Pushers Interactive LLC Page Cloud Bursting • Your data centre is maybe 5-20% utilised • You already have burst capacity • You haven’t implemented burst capability • Public Cloud is part of growth 42 http://blogs.barrons.com/techtraderdaily/2016/04/14/amazon-seeing-momentous-change-of-guard-as-public-cloud-booms-says-jp-morgan/
43. © 2015 Packet Pushers Interactive LLC Page SD-WAN is Old 43 • Viptela & Velocloud 2012
44. Selling Network Software Platforms not Products
45. © 2015 Packet Pushers Interactive LLC Page Software is Cheaper ? • Companies don’t/can’t sell cheaper products • Commoditisation means selling products • Three ways to go • Scale, Branding and Platforms 45
46. © 2015 Packet Pushers Interactive LLC Page Scale • Dell/EMC merger • get bigger • make it cheaper • increase market share -> branding 46
47. © 2015 Packet Pushers Interactive LLC Page Branding • Build a brand that customers want • Cisco, Microsoft, Oracle • Customers buy brand as long as product is good enough • requires a lot of marketing 47
48. © 2015 Packet Pushers Interactive LLC Page Platforms • Set of products & technologies that act in cohort to deliver services. • Interfaces and APIs connect subsystems. • Platform users do not see or know the detail of how the functionality provided by the platform is implemented. • Platforms commit customers to long term spending 48
49. © 2015 Packet Pushers Interactive LLC Page You want value 49 But you pay for products Sell you one of everything
50. © 2015 Packet Pushers Interactive LLC Page End of Products • You won’t buy a “router” or “firewall” • By a WAN platform • Or a security system • Data Centre Orchestration 50
51. © 2015 Packet Pushers Interactive LLC Page Platforms • Mesosphere / DCOS / Containers • Openstack / et al / Virtualisation 51
52. © 2015 Packet Pushers Interactive LLC Page Platform Quality • Lets be negative about things. 52
53. Analytics of Things
54. © 2015 Packet Pushers Interactive LLC Page Monitoring is Dead • The old model of FCAPS is dead • Pull, Polling (SNMP) • Log match triggers (Syslog) • Threshold Alerting (charting) 54 Fault Configuration Accounting Performance Security
55. © 2015 Packet Pushers Interactive LLC Page Analytics • Many Data Sources • Ingest into “Big Data” • Analysis / Machine Learning 55
56. © 2015 Packet Pushers Interactive LLC Page Cloud Analytics • Trigger Alert! • Using Big Data Analytics to mine collective data sources 56 Heuristics Ingestion Machine Learning Analytics & Telemetry Local Telemetry, Cloud-based Analysis On-Premises Off-Premises Web Interface Static Reports Cloud Storage Collection Agent Collection Agent Collection Agent Collection Agent Collection Agent Collection Agent Collection Agent Collection Agent Syslog Flow Data Windows Packet Capture FWL Logs IPS Logs Collection Agent Notifications/ Alerts © Greg Ferro 2015
57. General Issues
58. © 2015 Packet Pushers Interactive LLC Page Public Cloud is Small • Worldwide IT spending is forecast to total $3.49 trillion in 2016, a decline of 0.5 percent over 2015 spending of $3.5 trillion, according to Gartner, Inc • “Data center systems' spending is projected to reach $175 billion in 2016, a 2.1 percent increase from 2015” • “Amazon Web Services brings in $2.4B in revenue in Q4 2015, up 69% over last year" 58 Source: http://www.businesswire.com/news/home/20160407005953/en/Gartner-Worldwide-Spending-Forecast-Decline-0.5-Percent Source: http://venturebeat.com/2016/01/28/amazon-web-services-brings-in-2-4b-in-revenue-in-q4-2015-up-69-over-last-year/
59. © 2015 Packet Pushers Interactive LLC Page Public Cloud is “Cheap” • How many companies have bad organisation ? • move to the cloud to overhaul process and management • Stop spending capital • company has limited ability to finance itself (time to leave?) 59
60. © 2015 Packet Pushers Interactive LLC Page Partnerships • Seeing a lot of “partnerships” • Clear sign that no one knows what to do • partnering avoids commitment (especially cash) • highly vulnerable to market at large • acquisition (Aruba/HPE, Brocade/Ruckus) 60
61. © 2015 Packet Pushers Interactive LLC Page Startups • Buying from Startups is a complex issue • Agile, speed and changeable • which is good and bad • Picking startups is hard • BigSwitch Networks • Acquisition can be product death • Have best/better products • better support 61
62. © 2015 Packet Pushers Interactive LLC Page Sales Problems 62 http://blog.hubspot.com/sales/the-first-sales-call-conundrum
63. © 2015 Packet Pushers Interactive LLC Page One of Everything • Once upon a time, you chose a strategy which chose your vendor for you. • Now your vendors have all the strategies, you have to pick which strategy and which vendor 63
64. Business of Enterprise IT
65. © 2015 Packet Pushers Interactive LLC Page IT as a Luxury Market • Enterprise IT vendors have high gross profit margins typically in excess of 60% (up to 80%) • After costs of sales and operations, the net margin is 40-50%. • Raw materials cost is low - typically less than 10% of sale price. • Enterprise IT is a Luxury market • Chanel, Hermes, Ferrari etc 65 Source: http://richkidsofinstagram.tumblr.com
66. © 2015 Packet Pushers Interactive LLC Page First Class Assets Second Class Staff 66 • Has High priced technology delivered value ? • Do you even know ? • Does your Company use high priced assets anywhere else ? • Are we really sure that that paying extra delivers better results than buying cheap ? • Less money for Staff > lower quality staff • open source success stories Source: http://richkidsofinstagram.tumblr.com
67. © 2015 Packet Pushers Interactive LLC Page Museums vs Factories • Museum-grade Infrastructure • highly valuable • long lived, slow moving • careful, fragile • one of a kind • Industrial Infrastructure • low cost, • fast moving, short use/disposable • robust, rapid, • mass manufacture, volume, repeatable 67 Data Centre as a Factory Goods In - Goods Out. Fast, Value Added and Quality Industrial Production Goods In Services In Waste Products Services Out Compute Storage Network Operations Maintenance “Software Based” Machines
68. © 2015 Packet Pushers Interactive LLC Page Is Your Job Safe ? 68
69. © 2015 Packet Pushers Interactive LLC Page New Career Plans • Less specialisation, more generalisation • Constant learning / addition in narrow disciplines 69 Specialist or Generalist Yesterdays’ Model Specialist Breadth of Knowledge Depth of Expertise © Greg Ferro 2016 T-Shaped Todays’ Model Specialist Generalist Breadth of Knowledge Depth of Expertise © Greg Ferro 2016 Trident-Shaped Tomorrows’ Model Generalist Breadth of Knowledge Depth of Expertise © Greg Ferro 2016
70. © 2015 Packet Pushers Interactive LLC Page ITIL • ITIL was design to prevent failure. • It also prevents being successful. Or good. • ITIL is a reaction to bad habits of 1990s • UNIX sysadmins, bad attitudes, • Incompatible mini & micro-computers • Same for TOGAF, PRINCE2 etc • The time for ITIL is over 70
71. © 2015 Packet Pushers Interactive LLC Page Failure is Real 71 Source: Gartner (June 2012) urcre: https://thisiswhatgoodlookslike.com/2012/06/10/gartner-survey-shows-why-projects-fail/
72. © 2015 Packet Pushers Interactive LLC Page Attitude to Failure • Constant small failures are better than few big failures • But this is hard to understand in a pseudo- culture where ANY failure is unacceptable • ITIL rewards big failures • because no one can be blamed • “we did everything we could” 72
73. © 2015 Packet Pushers Interactive LLC Page IT Is Not A Team Sport • IT “teams” are modelled on sport or military • play a position • don’t play out of position. • single skilled, single focus • Future is Multi-skilled • know all positions • work in any team needed • deep in some, learn what you need 73
74. © 2015 Packet Pushers Interactive LLC Page Funding • Need / Gap / Report / Quote/Purchase / Implement • Cloud …. • Implement Project , Capacity Planning, Quote/Purchae • New methods of funding needed 74
75. © 2015 Packet Pushers Interactive LLC Page Quality • Product Quality will be a big issue • Would you buy and operate a car that needs a mechanic to install ? • Why do we expect first-gem product to • A paid-for product must be more reliable than open source 75
76. © 2015 Packet Pushers Interactive LLC Page Brands and Reliability • My current perspective • Whitebox has the same level of physical reliability • OEM SFPs have same level of 76
77. © 2015 Packet Pushers Interactive LLC Page The End Thanks 77
78. © 2015 Packet Pushers Interactive LLC Page Newsletters 78 Deep Dive Podcasts Latest News http://packetpushers.net/podcasts/ http://packetpushers.net/the-network-break/ http://packetpushers.net/free-newsletter-magazine-subscriptions/