June 12th, 2013 NYETM
On Wednesday, June 12, 2013, OLC attended New York Enterprise Tech Meetup held this time around at Microsoft. Four presenters were lined up for the meetup: 1010data, the leading provider of cloud-based analytics of big data; Typfesafe, provider of the most scalable software platform designed for computing architecture of the future; Gurobi, developed to help companies solve problems by providing the best optimization solver possible; and Microsoft’s Window Azure IaaS, a platform where customers can host Windows or Linux server’s workloads quickly and easily on enterprise-grade software at an extremely low price.
Typesafe was presented by Jonas Bonér, the CTO and co-founder, as “providing the world’s first reactive platform.” Typesafe has three core packaged products: Scala, Play! and Akka. It is designed to meet the needs of software engineering projects today. It has multiple verticals, with Bonér going as far as to say, “It’s a very generic product.”
Typesafe is trying to solve the way software is written. The way it’s being written is changing and the demands and expectations for applications have changed dramatically.
“In recent years, today’s apps are deployed on a wide range of infrastructure,” Bonér said. He listed the use of thousands of nodes and an uptime of 24/7 as examples of infrastructures. Typesafe provides new tools for a new era. “Developers need to build systems that react to events, react to load, react to failure, react through rich and engaging UI. Event-driven architecture means more effective and efficient resource utilization and management. To make this happen, you need to embrace asynchronous sending of events, non-blocking execution of events. Event-driven architecture is concurrent by design and fully reactive. It’s on-demand execution,” Bonér said.
Bonér talked about the scalability of Typesafe, explaining that the benefits include elasticity—scaling up and down. “It should be a deployment issue [regarding scalability], it shouldn’t be part of design. Topology and scalability is a deployment or runtime decision. This is cost-efficient hardware utilization,” he said. Scalability is reached because event-driven is the foundation for scalability. Its loose coupling and level of indirection makes location transparency possible. “Not transparent distributed computing makes the network explicit in programming models,” he said.
He talked about Typefast’s resilience next. “Resilience, to a large extent is ignored by the industry. If not completely ignored, it’s bolted on afterwards. We don’t do that with Typesafe. Our app self-heals. We build the product up in compartments to prevent cascading and failures. It provides full decoupling between business logic and error handling. It works in a distributed environment with the same semantics. The reactive traits form a cohesive whole and this needs to be future for cloud architecture,” Bonér said.
He briefly talked about Typesafe’s platform. It gets users started as a bootstrapping tool and can be used with Scala or Java. Once it is deployed. A rich real-time UX is presented and users can get started easily with building a reactive platform. Typefast uses an Apache 2 license for open-source. It monetizes through subscriptions and customer support.
1010data, based in New York, takes large amounts of data and makes it easily accessible for analyzation. 1010data has been in business for over 12 years and they have over 300 clients in their SaaS system. 1010data’s database allows their clients to analyze data regardless of what it is. It was revealed that the New York Stock Exchange was on of their first clients, accruing more than 1.5 trillion rows of data. The notion of analyzing data sets is what 1010data does. The speed at which the user moves is how fast the database responds. 1010data is opening up data, allowing users to perform anything with it. It performs as a database, analytics tool and much more. Computations are incredibly fast on 1010data and self-joins can be done in less than a minute even at 52 billion rows of data. Large leaps of what the user needs to analyze is possible.
1010data uses value expressions to search for data that is within the user’s question.
With just $150,000 worth of hardware, large bodies of data can be sifted through in mere minutes. Code can be generated behind queries as users search for more sets of data, definitions can easily be changed, data can be narrowed in scope or widened with no friction. Users can in fact compute arbitrary things.
1010data comes complete with a user’s manual to offer help on variables and functions to complete algorithms. Its code language is intuitive and easy-to-follow. The compiling code takes up to two minutes, depending on internet traffic.
Bob Hunt, a technical evangelist from Microsoft, demoed Windows Azure IaaS Virtual Machines. The Azure service provides IT professional experience, support for key server applications, easy storage manageability, high availability features, advanced networking and integration with computing PaaS.
“Microsoft is able to host workload wherever you like it to be,” Hunt said. He listed the physical, virtual, IaaS, PaaS and SaaS as places that can be hosted. “We’re going to be focusing on IaaS right now, as it introduces new features that allows full control and management of your platform. Users get access to a portal—navigation bars on the left side, tabs on the bottom [for easy maneuverability],” he said. Users can create a virtual machine very quickly. They also have the ability to choose where servers are located, what OS it can be run on, the size of the machine—and all are done within two minutes.
The visual machine can be provisioned in under 10 minutes and the size of the servers can be changed on the fly.
“For IaaS users, they have the ability to manage: App, data, runtime, middleware and OS; while Microsoft handles visualization, runtime, servers and more,” Hunt said. The Microsoft approach is to use a hybrid cloud, a broad and deep array of solutions that enable customers to use cloud in their own way.
“Microsoft Azure acts as a datacenter in the cloud while protecting private virtual networks in the cloud,” Hunt said.
Gurobi Optimization was presented by its CEO, Bob Bixby. Gurobi is pushing the frontiers of optimization performance and was founded in July of 2008. Its first product release was in May 2009 as a direct competitor to CPLEX. Gurobi is an optimization solver engine. It fits into big data and analytics, helping users make decisions about what is right for their business. “It’s a linear integer solver,” Bixby said. “As an engine, it’s what makes websites run efficiently.”
Gurobi was revealed to be a performance leader with outstanding support and no surprises. Its customers include Google, Microsoft, Walmart and FedEx. Gurobi is used from advertising to transportation to workforce scheduling. “Gurobi is horizontally scalable,” Bixby said.
He gave examples of Gurobi in use in these spaces:
• Workforce scheduling (McDonalds)—skill assignment & rostering
• Finance (BNY Mellon)—asset allocation, risk management, real-time trading
• Social network (eHarmony)—preference matching
• Transportation (USAir)—crew scheduling, fleet assignments
• Logistics (Walmart)—network, inventory, distribution planning
“Gurobi Optimizer is not personalized for any of these sectors—they all built on top of Gurobi,” Bixby revealed.
What Gurobi solves is mixed-integer programming, linear & quadratic programming and second-order cone programming. Its computing servers enable client-server apps. It allows users to seamlessly solve optimization models remotely. For Gurobi, performance is key. The current version is 16.4 times faster than its first version created in 2009.