As more organizations struggle to deal with the increasing volume, size, and complexity of today’s data, the traditional approach of storing data locally and then having an employee analyze it afterward is simply no longer a viable option. In order to stay competitive in a data-driven business world and to continue to support the growing business demands that come with this data increase, modern enterprises need to modernize their infrastructure and architecture. Fortunately, the advent and popularity of in-memory database technologies have made it possible for enterprises to improve data ingest, data processing, and data analysis in order to help companies make the best business decisions.
In its purest form, an in-memory database is a type of database that stores most (if not all) critical data in main memory (rather than traditional methods that store on a disk) to support various types of workloads and to enable faster processing and response times. This is not a new concept; however, it has been on the rise as of late because businesses are seeking newer and faster ways to get actionable insights and operational analytics. This demand for real-time data and real-time results is what will make or break a business in a market where having a complete customer profile is imperative to providing superb customer experience to help garner customer loyalty.
SourceForge spoke with Manish Gupta, the Chief Marketing Officer of Redis, about the importance of real-time data and how Redis delivers the performance, flexibility, and scalability businesses need. Gupta also shares some insights about the emerging trends in the enterprise database market.
Q: Please tell us more about Redis. When did the company start and what problems do your solutions seek to address?
A: Redis was founded in 2011 and is headquartered in Mountain View, California. The development center, meanwhile, is located in Israel, where two of our co-founders are located.
The idea behind the founding of Redis is this: mobile applications, IoT, and social media collaboration generate a massive amount of unstructured data for organizations to manage. This data does not fit neatly in strict columns and rows. Enterprises need an alternative model for database management–which is why non-relational databases have generated a lot of enthusiasm in recent years.
Manish Gupta, Chief Marketing Officer at Redis
Non-relational databases can take data in any structure/format: pictures, streams, documents, etc. at any speed and can store it. And when the system is reading over this information from the database, we can request retrieve the data in a particular format, which offers ultimate in flexibility for folks who are developing apps.
Redis leveraged the open source Redis project, enhanced it, then commercialized it. This became the enterprise version known as Redis Enterprise. There was a massive movement in the industry where all of a sudden, unstructured data volume was exceeding structured data volume. High velocity unstructured data is challenging for enterprises to leverage as it becomes too complex to organize and understand…that’s where Redis comes in.
We started with a cloud offering but also offer the ability to run Redis Enterprise as downloadable software on-premises or in a hybrid environment (where some data might be in the cloud and rest on-premises). Today, Redis has over 70 thousand accounts worldwide and 7500 enterprises that are paying customers.
Q: Data is now viewed as a strategic asset and the fuel of the digital economy. How can enterprises unleash the power of data?
A: In today’s marketplace, more data is traversing the enterprise, but also the speed at which this is happening is increasing so the volume is nearly impossible to leverage in the enterprise. More data is better so that you can get better insights, but unless you can ingest, clean the data, process the data, analyze the data, and then present insights from it, it is not doing you any good. In fact, all of that data creates incredible level of complexity.
Historically, only about 12% of all data has been touched by enterprises; a large percentage of data simply goes untouched. We are now layering in machine learning and Artificial Intelligence (AI) where human intervention is extracted away from the process of churning data. Machines and bots are doing much of the assessment and presenting a thin sliver of the data to provide a way for humans to interact in the process. Much of the decision making is happening at the automated level because it’s the only way for enterprises to handle all of the data coming in.
Q: How will certain advancements such as Artificial Intelligence and Machine Learning enable enterprises to intelligently manage their data?
A: Leading analyst firm, Gartner predicts that by 2020, 95% of image- and video-oriented content will never be seen by humans. Instead, machines that provide some degree of automated analysis will have the ability to check and examine it.
The reason behind this is that too much data is coming in. Humans have the ability to take advantage of this data (e.g. processing for marketing campaigns) but trying to analyze all of the social media streams, text, images, and video is difficult. If a human is analyzing this data off of prospects it becomes simply impossible due to velocity. We provide machine learning algorithms that collect categories of data and present it to a marketing manager so that they can say “based on this data from the past few days, prospects in this category are more likely to buy.” You need that layer of machine learning in order to add that additional insight and layer.
Another thing is that, as a society, we are used to hitting or clicking buttons on our mobile device, and we are looking for that response and response time to be hyper-customized to what we want, when we want it. Just take Uber as an example: a three minute wait time might be too much for a consumer. In the backend of Uber, there is a metric that is attempting to deliver a hyper-customized response (cost, time, ETA, etc.) in milliseconds. Take a second to think about the number of operations that need to happen in the background: location, assessment of cars nearby to map to you, calculation of time, traffic between you and your destination, map to dollar value, etc…all of this has to happen in a matter of milliseconds.
In essence, historical data mapped to real-time must be matched together in a matter of seconds. Our expectations are for a hyper-customized experience in the shortest time possible. This has to happen where you are, too. Plus, we are finding that analytics is running on the edge of the network instead of in the cloud. The databases supporting these applications have to respond in real time. An Oracle SQL database for example, has almost no play in this kind of real-time experience.
Q: Redis is lauded as the world’s fastest database. How does Redis enable enterprises to cost-effectively meet their real-time data needs?
A: Redis’ Redis Enterprise database is non-relational, runs in-memory, can operate in public clouds, on customer’s VPC, on-premises as well as in hybrid manner. Due to its efficiency, it requires very few compute resources and is highly efficient. This results in infrastructure savings for the customer.
Recently, we launched Redis on Flash, the ability to balance workload between RAM and SSDs where you can keep key and hot value sin RAM and cold values in SSDs. This intelligent memory tiered approach can offer dramatic cost savings without taking a significant performance hit.
And finally, simplicity of Redis and availability of a large community reduces human resource costs when running Redis Enterprise databases to support modern applications.
Q: Please tell us more Redis Enterprise. How does it enable enterprises to meet their real-time data needs?
A: The hallmark of Redis Enterprises is its high performance and the ability to run efficiently. With it, we are able to run a million operations (write or read data from database) per second with less than one millisecond of latency on just two moderately powered servers.
Redis Enterprise enables enterprises to achieve quick time to market. The Redis Enterprise database incorporates a variety of data structures that can be stitched together, like Lego blocks, to deliver complex functionality in a simple manner. This is very powerful when application developers have to enable large amounts of functions while giving intuitive user experience.
The versatility and extensibility of Redis Enterprise also make it a stand out in the field. You can use it for diverse real-time use cases and solutions ranging including IoT, personalization, eCommerce, fraud mitigation social and metering.
Q: Redis recently raised $44 million in funding. What can enterprises expect from Redis in the future?
A: At Redis, we pride ourselves on our ability to deliver accelerated innovation. Recently, we announced a streaming capability; the enterprise platform also revealed integrated search capabilities. In the coming months, you will see an increasing amount of native, integrated functionality within the platform.
Right now we working closely with some of the ecosystem players in the analytics, IoT, as well as in machine learning domains.
Q: Looking ahead, what rising trends, challenges, and technologies do you think will transform the enterprise database market?
A: One of the biggest trends I’m seeing is the evolving definition of the term “real-time.” We’ve been talking about real-time forever and the definition has changed now to a point where real-time means instant. In business today, the official definition of real-time is completing an operation within 15 minutes; this window is getting smaller and smaller every day. Consumers, on the other hand, expect their demands to be met in a hyper-customized manner instantly. Databases have to deal with read and write operations within milliseconds.
The second trend would be the need of modern applications to handle hybrid workloads: analytics and transactions combined and running on the same dataset. Databases need to have the capability to support these hybrid operations which demand speed, volume, and data structure flexibility.
The third trend (which is more at a system level) is the tight integration of databases with the environment in which they operate. Applications will run in the cloud, on-premises, and some in a hybrid fashion. Databases have to be architecturally flexible to have the ability to run in a “native” mode in all three deployment scenarios.
About Redis
Headquartered in Mountain View California, Redis is the home of open-source in-memory database platform Redis and provider of Redis Enterprise. The Redis Enterprise platform, an enhanced version of open source Redis, delivers unmatched resilience and availability, scalability, and cost reduction. The deployment options for Redis Enterprise include Redis Cloud, Redis Cloud Private, Redis Pack, and Redis Pack Managed. Redis is trusted by over 7,500 enterprise customers worldwide.