Prioritizing Flexibility in the Cloud: A Tracky Success Story

With the global public cloud market expected to reach revenues of $66 billion by 2016, it’s clear that businesses of all sizes and verticals are turning to the cloud for its flexibility, accessibility and scalability benefits. But, for SMBs with limited staff and resources, migrating to the cloud often poses challenges.

After battling restrictive providers’ fixed server instances, bundled resources, high costs and limited deployment options, Web-based social collaboration company Tracky knew it needed to prioritize finding a more flexible public cloud provider – one that provided the high performance computing environment and seamless experience its growing user base required.

In this session, Tracky Vice President of Development David Longnecker and CloudSigma CEO Robert Jenkins will take a deeper look into Tracky’'s migration to public cloud, explaining why a flexible cloud provider is a necessity for growing SMBs that require a robust, high performance computing environment, but won’t break the bank.

Transcript

2. SMBs Have Unique Challenges
3. TRACKY – FINDING FLEXIBILITY Connect. Collaborate. Share.
4. Tracky Company Snapshot • Founded in 2011. • A “social collaboration” platform connecting you, your social networks, and the things you need to do—all in one place. • A “social publishing” platform building online communities through collaboration.
5. Tracky’s Story • Small, rapidly growing SaaS platform. • Frequently changing needs. • Physical hardware too restrictive, time consuming.
6. Is this your server environment? Tim Berners-Lee’s original NEXT workstation for developing the WWW
7. Tracky’s Story What do we have to gain?
8. Tracky’s Story What do we have to gain? What do our customers have to gain?
9. Tracky’s Story Finding the right provider can be really hard!
10. The Unknown Problems SMBs with physical infrastructures may not anticipate “the lacks”. • Lack of flexibility • Lack of configurability • Lack of performance
11. Flexibility Small, medium, large… jumbo?
12. Flexibility We needed flexibility! Provider s c a l eis important. – too expensive to grow into – wrong mix of features – inefficient
13. Configurability We needed control! - Our environment - Our rules
14. Performance We needed speed! First impressions of a SaaS is key!
15. What can you sell your customers in a quarter second?
16. What Does Speed Cost? +500ms  -20% traffic @ Google +100ms  -1% sales @ Amazon Source: presentation by G. Lindem, Amazon: “Make data useful”
17. CLOUDSIGMA – A CLOUD OF OPPORTUNITIES freedom through technology
18. CloudSigma Company Snapshot • Founded 2009, first cloud live Q1 2010 • 500% revenue growth year over year 2010 – Q1 2013 • Offices in San Francisco, Zurich, Sofia and representation in New York and Lima • Approximately 40 employees • Three cloud locations: Las Vegas, Zurich1 and Zurich2 • Additional locations in progress: Washington DC, Amsterdam, Sao Paulo, London & Singapore
19. The Problem in IaaS… ..is generic virtual machine sizes trying to fit all compute problems in one size, leading to waste
20. The Solution is… • Utility billing of unbundled compute resources leading to efficient purchasing • Direct private connect hybrid cloud extension – capture enterprises • SSD priced as magnetic disk – no contention or iowait • And many other values we deliver
21. The Product • Unbundled Compute power Infrastructure- as-a-Service • Compute power in the form of Virtual Machines • Rented on 5-minute billing cycles
22. Some of our Customers & Partners
23. Product – Market Fit
24. Plan Do Check Act Everything we do is iterative…
25. Time Product delivered To avoid any self delusion…
26. A primary challenge of public cloud is delivering predictable performance levels based on unpredictable workloads.
27. Problem: Networking bottlenecks
28. Solution: Dual 10 GigE/40 GigE and Software defined networking
29. Problem: Variable storage performance
30. Solution: All SSD Storage
31. Problem: CPU over allocation leads to unpredictable performance
32. Solution: Resource allocation becomes load balancer with live migration
33. Full performance SLA SDN Delivering Guaranteed Networking Metrics SSD Delivering Guaranteed Storage Performance Resource Allocation & Live Migration Delivering Stable Compute Performance Customer Defined Computing
34. Cloud Ecosystem advantage: Increased Choice
35. Dealing with failure • S**t happens • You need to be prepared to deal with it • The cloud has failures just like all other IT systems • The cloud offers ways to abstract to avoid such failures more easily You need to be given the tools to achieve high availability and know how to use them!
36. BAD! Physical compute node 1 Physical compute node 2 Database VM node A Database VM node B Storage System 1 Virtual drives Virtual drives VRRP failover Achieving separation
37. GOOD! Physical compute node 1 Physical compute node 2 Database VM node A Database VM node B Storage System 1 Storage System 2 Virtual drives Virtual drives VRRP failover Achieving separation
38. Storage System 1 Achieving separation Physical compute node 1 Physical compute node 2 Database VM node A Database VM node B Storage System 2 Virtual drives Virtual drives VRRP failover I’m down!!
39. Achieving separation Physical compute node 1 Physical compute node 2 Database VM node A Database VM node B Storage System 1 Storage System 2 Virtual drives Virtual drives VRRP failover I’m down!!
40. What to look for • Transparency over computing and storage resource allocation • Tools to express resource separation requirements • Support for multicast traffic & other high availability system needs
41. Questions? David Longnecker Tracky davidl@tracky.com @dlongnecker Robert Jenkins CloudSigma robert@cloudsigma.com @cloudsigma