2. Alex Hübner - Who?
n Software Manager, leader for SDx initiatives at NubeliU & Logicalis Brasil
n Previously: Embratel (AMX), Inter-American Development Bank (IDB),
UOL Diveo, Locaweb and my own solo-consulting company
n Father of two little angels: Corina (6) and Oliver (2)... (living the "wild" life) J
n Studied Geology, learned how to fly airplanes
(yes, with a brevet) but ended-up "working
with computers"... (as my family got used to describe)
n Climbed Mt. Pisco in 1997. Almost died in
Huascarán in 1998!... Perú is one of the most
beautiful contries in the World!
3. NubeliU Overview & Key People
Founded in 2014, with offices in Argentina (Buenos Aires), Brazil (São Paulo) and
partners all over the world, NubeliU sucessfully builds, manage and optimize
software defined infrastructures for multiple cloud workloads.
Rodrigo Benzaquen
Co-founder & CEO
Former Site Operations & Infraestructure
Director at MercadoLibre, where he was
the first technology employee and
worked for14 years.
Richard Hager
Co-founder & COO
Former executive and consultant for
several brazilian internet companies. Early
technology employee at MercadoLibre.
Alejandro Comisario
CTO
Former Cloud Infraestructure TechLead at
MercadoLibre, where hewas one of the
founders of theMercadoLibre’s Private
Cloud.
Maximiliano Venesio
Chief Cloud Architect
Former Cloud Storage Tech Lead at
MercadoLibre.
Leandro Reox
VP of Engineering
Former Networking Virtualization Tech
Lead at MercadoLibre.
4. NubeliU Expertise and Excellence
Professional Services
• Cloud Architecture & Design, focused on OpenStack & IaaS
• Web Application Performance & Scale
• Infrastructure Cloud Building
• Application Cloud Migration
1
Managed Services
• Outsourcing of cloud management
• 24x7 Infrastructure, Applications and Business Monitoring
• Software & Infra DevOps Integration
2
Cloud Economic Analytics
• SaaS or On-Premise platform for highly scalable Multi-Cloud Analytics
• IaaS, PaaS e SaaS multi vendor cost monitoring
• Cloud Infrastructure Optimization
3
6. NubeliU's "Breakfast of Champions"
NSO
ESC
YANG
and others...
OR
An example of the type of work (and complexity) we're now seeing in the world. A
work that requires a different system integration approach.
Orchestration and Management Stack for a pure "TelcoCloud" deployment:
Cisco FocusLogicalis/Nubeliu Focus
13. It all started in "Software is eating the World"
"Why Software is Eating The World"
by Marc Andreessen @ WSJ - August 2011
https://goo.gl/bFswZv
Marc Andreessen article for the Wall Street Journal in August 2011
14. Definitely a trendy topic...
Oct, 2017May, 2015Oct, 2013
Interest over time:
15. Looks like a hype,
sounds like a hype,
it must be a hype...
?IT IS
16. Digital Transformation, real or not?
It doesn't matter if it's a hype or not. All you need is look
around. You'll notice things are really changing (really fast).
Shaw we look to some evidences?
17. Digital Transformation, a closer look:
Source: CBInsigths, August 2016
Companies with a valuation of over U$ 1 billion in the last 5 years:
18. Digital Transformation, a closer look:
Source: CBInsigths, August 2016
Companies with a valuation of over U$ 1 billion in the last 5 years:
19. Digital Transformation, a closer look:
Source: CBInsigths, August 2016
Companies with a valuation of over U$ 1 billion in the last 5 years:
20. Digital Transformation, a closer look:
Years to reach a U$ 1 billion valuation. The "typicals" versus the "unicorns":
Source: World Economic Forum White Paper Digital Transformation of Industries – Jan 2016
21. Digital Transformation, a closer look:
Years to reach a U$ 1 billion valuation. The "typicals" versus the "unicorns":
Typical Fortune 500: 20 years
Source: World Economic Forum White Paper Digital Transformation of Industries – Jan 2016
22. Digital Transformation, a closer look:
Years to reach a U$ 1 billion valuation. The "typicals" versus the "unicorns":
Typical Fortune 500: 20 years
Source: World Economic Forum White Paper Digital Transformation of Industries – Jan 2016
Average "Unicorns": 4.4 years
23. Digital Transformation, a closer look:
Years to reach a U$ 1 billion valuation. The "typicals" versus the "unicorns":
Typical Fortune 500: 20 years
Source: World Economic Forum White Paper Digital Transformation of Industries – Jan 2016
Average "Unicorns": 4.4 years
24. Digital Transformation, a closer look:
Years to reach a U$ 1 billion valuation. The "typicals" versus the "unicorns":
Typical Fortune 500: 20 years
Ops... 52% of the companies
apearing in the "Fortune 500" in the
year 2000 are now gone.
Source: World Economic Forum White Paper Digital Transformation of Industries – Jan 2016
Average "Unicorns": 4.4 years
25. Digital Transformation - Conditions
What/Why is different now?
Six decades into the computer revolution, four decades since
the invention of the microprocessor, and two decades into
the rise of the modern Internet, all of the technology
required to transform industries through
software finally works and can be widely delivered at
global scale.
Marc Andreessen, 2011
" "
26. Digital Transformation - Conditions
There are at least 4 billion people connected to the Internet at
decent speed (+5 Mbps) world-wide. In 2001, at the high of the
.com bubble, we were less than 100 million...
We now have Internet
(REAL Internet)
Source: Fixed (wired)-broadband subscriptions per 100 inhabitants 2012, Dynamic Report, ITU ITC EYE
27. Digital Transformation - Conditions
Today, more than 2.5 billion people carry a real computer in its
pockets. These "computers" are equipped with cameras, moving
sensors, GPS and 24x7 Internet connectivity.
We now have
Smartphones
Source: Active mobile-broadband subscriptions per 100 inhabitants 2012, Dynamic Report, ITU ITC EYE
28. Digital Transformation - Conditions
Today anyone has virtually infinite computing and storage capacity
without having to spend a penny on CAPEX and/or maintain
complex infrastructure and operations.
We now have
Cloud Computing
29. Digital Transformation - Conditions
Today we no longer need to worry about marginal aspects of our
business. From a simple database of ZIP codes to a sophisticated
correlation processing analysis in images through cartography,
geoprocessing, etc., there's a platform to run that!
We now have APIs and
Platforms for pretty much
everything!
31. Cloud adoption is reaching new hights
Public & Private Clouds are quickly approaching Traditional Data Center
spenditures, most likely surpassing by 2020.
Source: IDC Worldwide Quarterly Cloud IT Infrastructure Tracker. Q3 2017
Cloud & DC Infrastructure Spend, Global, 2015-2021
100%
80%
60%
40%
20%
58,66%
25,09%
16,25%
48,59%
31,00%
21,41%
32. Cloud concerns are changing
Concerns are changing to reflect a new reality, one where multiple clouds cohexist,
serving multiples workloads and purposes.
Source: Bain Cloud Computing Survey, 2015; Morgan Stanley AlphaWise Survey of IT Managers
Top 3 Cloud Concern, USA, 2012-2015
"GETTING OLD" "NEW" "NEW"
33. Hybrid-Cloud has won, period.
58% of today's enterprises have a "Hybrid-First" cloud strategy.
Source: RightScale 2017 State of the Cloud Report
Enterprise Cloud Adoption Model (respondents with 1K+ employees)
34. Hybrid-Cloud has won, period.
Hybrid-Cloud became the dominant model. Multi-Cloud is the new thing.
Cloud type adoption, Brazil, 2013-2016
Source: Brazil IT Snapshot 2017, Logicalis
37. Hybrid and Multi-Cloud are the same?
Yes and no. Nevertheless, it's important to highlight some of the differences/nuances:
n Hybrid-Cloud: on-premisses (private) + external (public) cloud with data exchange
between those. Main drivers are technical (performance, compliance, licensing, etc.);
HYBRID-CLOUD MULTI-CLOUD
n Multi-Cloud: any combination of cloud types. Companies might opt to use more than
one private cloud; combine several public clouds; and mix-up this with hybrid clouds
too. Data exchange is not mandatory. Main drivers are budgetary (cost) and strategic
(avoid lock-in).
40. Open Source for Hardware too!
The physical future of open source.
Supported by
OPEN
ComputeProject
41. Open Source for Hardware too!
The physical future of open source.
Supported by
OPEN
ComputeProject
Cisco joined the project in 2014
and it's currently a GOLD member
42. Open Source for Hardware too
The physical future of open source.
OPEN
ComputeProject
Before entering ...
Visiting Facebook's Prineville
DC in Oregon - July 2016
Special thanks to
43. Open Source for Hardware too
The physical future of open source.
I thought I knew something
about modern Data Centers.
Obviously I was wrong...
After leaving...
44. Open Source for Hardware too
The physical future of open source. Some interesting examples:
Facebook
"SIX-PACK"
2015
1.28 Tbps on each dataplane module, 40Gbps+ on each port...
45. Open Source for Hardware too
The physical future of open source. Some interesting examples:
Facebook
"BACKPACK"
2016
3.20 Tbps on each dataplane module,
100Gbps+ on each port...
46. Open Source for Hardware too
The physical future of open source. Some interesting examples:
Facebook
"WEDGE"
Wedge 40G - 2015
Wedge 100G - 2016
ToR (top of rack) or "Leaf"
switches with 40 Gbps &
100 Gbps ports.
47. Open Source for Hardware too
The physical future of open source. Some interesting examples:
Global Transponder DWDM
12x 100G QSFP28
04x 200G DWDM
But aren't these lab rats?... Well, better think again:
• Telia Company in Voyager tests with Facebook - https://goo.gl/YjfUKY
• Equinix Begins Trials of Facebook’s White Box Transponders - https://goo.gl/1NHy8u
• MTN tests Facebook’s Voyager platform - https://goo.gl/xJwwu3
Facebook
"VOYAGER"
2016
48. Open Source for Hardware too
The physical future of open source. Some interesting examples:
Facebook
"Bryce Canyon"
2017
Up to 72x HDDs - 3.5" SAS/SATA 12/6 Gb in a 4OU (equivalent of 4 RU) powered
by Intel Mono Lake CPU, sporting 50+ Gbit/s NIC in a 100% tooless design chassi.
Can store up to 720 TERABYTES per node (as of 2017).
JBODstorage
49. Open Source for Hardware too
The physical future of open source. Some interesting examples:
3.2Tbps (32 by 100G) platform that can be used as a leaf
or a spine switch.
"Pigeon"
2016
4U server with 60x 3.5"SAS/SATA disks
totaling 480TB of data storage that could
cost less than a nickel (U$0.05) per gigabyte.
"Storage Pod 6.0"
2016
50. Open Source for Hardware too
The physical future of open source. Open Compute Project (OCP) is not alone.
52. OpenStack is now mature
OpenStack is now "boring" and continues to grow steadily (44% CAGR YoY).
n Numbers, numbers and numbers... lots of numbers! (next slides).
n Continued focus on paving the way for telco transformation, 5G and IOT.
n But at the same time maintains a strong (and growing) presence in the
traditional IaaS/SDDC workloads;
n OpenStack its no longer only about "the cloud", its all about SDI/SDx!
§ That's why OpenStack continues extremely complex and hard to do (install,
configure, manage, maintain, upgrade and sometimes, understand)... :-)
n Big vendors and players continue to jump in the bandwagon:
§ Huawei and Ericsson are now Platinum members - platinum membershiping is not a
badging thing, is serious commitment!
§ Oracle is now advocating OpenStack as the main underlying platform for its products;
§ Google joined the project in 2015.
53. OpenStack is now mature
OpenStack Market Size, Global, 2015-2020.
Source: OpenStack Pulse Report 2016 – 451 Research.
54. OpenStack is now mature
Which industries use OpenStack?
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
55. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
Where in the world are the OpenStack users?
56. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
What size organizations use OpenStack?
unknown 2%
57. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
Why organizations choose OpenStack?
58. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
Which emerging technologies interest OpenStack users?
59. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
What types of clouds are running OpenStack?
60. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
Which workloads and frameworks are running on OpenStack?
61. OpenStack is now mature
Source: OpenStack User Survey 2017 (April Edition), OpenStack Foundation
Which OpenStack-powered solutions (distros) are most used?
62. OpenStack is now mature
Source: OpenStack User Survey 2017, OpenStack Foundation
Traditional IaaS/SDDC workloads. Notable/relevant examples around the world:
n SAP HANA
n Walmart
n PayPal
n American Express
n Time Warner
n Mercado Libre
n Comcast
n Bloomberg
n Rackspace
n Deutsche Telekom (IT cloud)
n China Telecom (IT cloud)
n UOL
n TIVIT
n Locaweb
n Equinix
n OVH
n Totvs
n Internap
n Wells Fargo
n Bank of America
n Itaú
n Santander
n Capital One
n Embratel
n Barclays
n Volkswagen
n BMW
n Nike
n (the list goes on...)
63. OpenStack is now mature
Distro consolidation, a clear sign of maturity (not weakness as some might think)
n Red Hat
n Canonical (Ubuntu)
n Mirantis (pivoting strategy)
n Huawei
n SUSE
n VMWare (VIO)
n HP Helium
n Oracle
n Platform9
n Cloudbase
n ZTE
n Others
"Waking the future": "Walking the plank":
Source: Hübner Research & Rumors Institute
2019 update...
64. An OpenStack love history...
For serious Clouds (and peace of mind):
And we also love:
VIO
66. Multi-Cloud Chaos
n The multi-cloud world we now live in is a result of
shadow IT and bimodal IT gone wild.
n Multi-Cloud tools (CMPs) are on its infancy. Cloud
targets are moving all the time. Hard to keep the pace.
n No one was really expecting it. No one has predicted
it the way it’s happening.
n The proliferation of cloud environments is increasing
very quickly, in some cases without the realization of
the extent to which it was happening. That’s how we
get IT chaos.
n How do we manage the things that have been (and
always will be) important? Things like risk
management, compliance, governance and cost
control. How do we bring order to something that is
often, as of now, completely out of control in many
organizations?
Some hard truths about Multi-Cloud:
67. Multi-Cloud Chaos
n Risk: Multiple cloud platforms means a broader attack surface with new
vulnerabilities, requiring new tools to maintain effective security, governance
and compliance in hybrid environments.
n Complexity: IT has traditionally focused on managing dynamic resources in
its own data center. Now it must keep a keen eye on any combination of
cloud platforms with different processes and capabilities.
n Speed: Complexity can slow down migration to the cloud, but in multi-cloud
environments, slow translates to broken. Enterprise users want immediate
availability and unlimited scalability from cloud-based services and will
circumvent IT to source public clouds themselves if their needs are not met.
n Cost: Businesses migrate to the cloud (public more significantly) with
expectations of realizing significant cost savings, but they are often met with
varying degrees of bill shock. Unrestrained service acquisition of public
cloud solutions can result in costs spiraling out of control while underutilized
or unused resources remain active.
Multi-Cloud complexities (just scratching the surface):
69. Multi-Cloud Chaos, Managed
n Proactively monitor cloud and infrastructure applications: Leverage
orchestration, automation, machine learning and analytics to quickly
and proactively detect and diagnose potential issues.
n Dynamically optimize multi-cloud capacity: Align IT resources with
business demand and identify infrastructure needs to improve
application performance and reduce costs when migrating to the cloud.
n Effectively manage the cost of multiple-cloud suppliers: Gain
insights into both on-premise and public cloud costs to enable
informed decisions between capital or operational expenses and
reduce bill shock.
n Accelerate access to multi-cloud resources: Enable policy-based
access to multi-cloud resources via a consumer-like self-service
catalog that improves governance and automates provisioning and
management.
Things to watch in a Multi-Cloud Strategy:
70. Multi-Cloud Chaos, Managed
n Increase visibility of infrastructure in the cloud: Discover multi-cloud
infrastructure and services you may not be aware of in order to
understand the complexity and dependencies of applications;
n Prioritize digital business automation: Reduce complexities around
the move to the cloud to achieve the desired performance, scalability
and cost savings. Automation can make this much less painful;
n Increase regulatory compliance automation: Identify regulatory
compliance issues and then prioritize, plan and accelerate automated
remediation to simplify and reduce the cost of compliance.
Things to watch in a Multi-Cloud Strategy (continued):
71. Brutal Automation is needed
Source: http://www.lightreading.com/automation/dt-brutal-automation-is-only-way-to-succeed/d/d-id/737111
LightReading,10/10/2017
Automation is key. You can't avoid. Better start working now.
72. Containerization is now a reality
Containers has a bright future. It’s no longer an adventure. Early adoption mode
is on (and not just for Dev purposes). Real workloads are being pushed.
73. Serverless Computing, demystified
Adrian Cockcroft
VP Cloud Architecture
Strategy at AWS
Serverless computing is a cloud computing execution model in which
the cloud provider dynamically manages the allocation of machine
resources. Pricing is based on the actual amount of resources
consumed by an application, rather than on pre-purchased units of
capacity. It is a form of utility computing.
A MORE FORMAL DEFINITION COULD BE:
" "