Tastewise: Finding the perfect recipe for faster, more efficient food-data search
Meet Tastewise
Trusted by leaders in the food and beverage industry, including Campbells, KraftHeinz, Nestlรฉ, Mars, and PepsiCo, Tastewise is a GenAI-powered food and beverage platform that transforms consumer data into unique insights. This allows food and drink brands to make decisions and execute at speed and scale.
Tastewise gives food brands fast access to data insights into emerging food trends and the GenAI tools to develop concepts, design products, and plan marketing and sales campaigns that cater to consumersโ changing tastes.
The Challenge
Food trends move fast. Products can go from obscurity to ubiquity almost overnight. For those catering to these fast-changing tastes, it can be hard to keep up. Gone are the days when a simple phone survey would do โ Food brands require extensive data to truly understand their customers and innovative tools to efficiently scale their marketing efforts. Thatโs where Tastewise comes in.ย
By gathering and storing a treasure trove of publicly available data, it gives brands instant insights and execution tools to keep pace with emerging trends and stay ahead of the competition.ย
Tastewiseโs cloud infrastructure runs mostly on Amazon Web Services (AWS), with its wealth of data stored in S3 buckets. Elasticsearch powers its search and analytics engine. However, the sheer volume of data Tastewise handles meant its storage costs were rising fast and challenging predictable unit economics. โUsually S3 isnโt an issue because itโs not that expensive,โ says Doron Gill, VP of R&D at Tastewise. โBut when you start getting into huge amounts of data, costs can start rising very quickly.โ Tastewise needed a more effective way of storing its data to maintain performance while keeping costs predictable at scale.ย
At the same time, the complexity of certain customer searches was placing a strain on Tastewiseโs system, impacting response times and risking timeouts against SLOs. โin extreme situations when users ask heavy queries, the Elasticsearch index is huge and it quickly becomes a very heavy operation,โ Gill explains. Tastewise needed an Elasticsearch expert to help it optimize performance to deliver fast insights with no timeouts.ย
Rather than spending valuable resources hiring in-house, Tastewise chose a product-led, partner-delivered approach to solve these challenges. It turned to DoiT and BigData Boutique.ย
The Solution
Keeping cloud costs in check with DoiT consultancy
Tastewise started working with DoiT in 2021 to optimize cloud economics. A key priority was to optimize Amazon S3 storage economics while preserving access latency and durability. DoiT recommended tiered storage to ensure Tastewise had fast access to the data it needed frequently while moving backup data and secondary files to more cost-efficient storage layers.ย
With Tastewise paying for a large number of API calls to access its stored data, DoiT also recommended right-sizing objects and batching access to limit the number of calls it had to make.
The DoiT team continues to analyze Tastewiseโs cloud usage via DoiT Cloud Intelligenceโข during quarterly reviews and makes recommendations to increase efficiency and strengthen cost governance.
For example, DoiT recently noticed Tastewise was running ElastiCache on on-demand instances and recommended switching to a commitment strategy (reserved). This simple change saved Tastewise $1,000 a month.
Simplifying FinOps with DoiTย
As well as relying on DoiTโs cloud expertise to help reduce its cloud bill, Tastewise also saves money using DoiTโs intelligent technology. With Flexsave for Computeโข, which automatically finds and applies discounted rates to Tastewiseโs EC2 workloads, Tastewise has saved money on its compute instances each month.ย
Using DoiTโs analytics technology, Tastewise gets a single, shared, real-time view of its AWS costs and the few services it runs on Azure and Google Cloud. โWith DoiTโs technology, I can see all our expenditure in one place, and I donโt have to go to three different platforms to work out our total costs,โ explains Gill.
With Real-Time Anomaly Detection, Tastewise receives alerts whenever its cloud costs rise unexpectedly, preventing month-end surprises and enabling immediate remediation. For example, a recent alert revealed that the company was spending more than expected on its CloudTrail logs.ย
This prompted Gill to investigate, discovering that the debug flag had accidentally been left on. Without Anomaly Detection, Tastewise would have had to wait until the end of the month to spot the issue in its monthly cloud bill. Instead, it fixed the issue immediately, saving the company thousands of dollars.ย
โI use Anomaly Detection all the time,โ says Gill. โSometimes the amounts are quite small, making them easy to ignore, but those amounts can quickly add up to a thousand dollars or more. DoiT shows us the anomaly as it happens, which helps us a lot.โ
Giving customers faster, smarter insights with BigData Boutique
With its costs under control, Tastewise needed specialist Elasticsearch support to deliver fast insights with uptime confidence. DoiT recommended its long-time partner BigData Boutique to Tastewise, and the companies began working together in 2022.ย
BigData Boutique initially assisted Tastewise in migrating from single-node Elasticsearch clusters to a more resilient, scalable infrastructure on AWS. Tastewise has gone on to leverage BigData Boutiqueโs Elasticsearch expertise for everything from routine maintenance to fixing performance issues as they arise, helping to give food brands the data insights they need on demand, without timeouts.ย
โElasticsearch is a great persistence engine,โ says Gill. โBut things like adding more nodes when our indexes are filling up, or rebalancing shards, require expertise, which BigData Boutique gives us. And if we ever experience downtime, BigData Boutique helps us identify the root cause. This is very important because, without our Elasticsearch up and running, weโre dead in the water.โ
Re-architecting data services with an Elasticsearch expertย
Tastewise and BigDataBoutique have also worked together on re-architecting its data stack, such as how the company ingests and processes data. Previously, data ingestion pipelines and customer queries occurred in the same compute, sometimes affecting customersโ ability to gain fast insights.ย
Recommending that Tastewise separate the data development and production environments, BigData Boutique helped to ensure the vast amount of data it ingests didnโt affect the stability of its customer-facing service.
Checking the health of Tastewiseโs data stack with Pulse
In addition to relying on BigData Boutiqueโs Elasticsearch expertise, Tastewise was an early adopter of BigData Boutiqueโs Pulse solution. With proactive monitoring, Pulse alerts Tastewise to performance issues causing slow queries. The health assessment feature of Pulse continuously analyzes the health of Tastewiseโs Elasticsearch clusters, and quickly pinpoints the cause of any issues.ย
With this information, Tastewise can then turn to BigData Boutiqueโs support experts for help fixing the problem, ensuring its search clusters are stable and providing a reliable service to customers.ย
โPulse is an excellent Elasticsearch observability platform,โ says Gill. โAnd the best thing is, it sits in our cloud infrastructure, so itโs not something our engineers need to maintain. All we need to do is show the insights to BigData Boutique, which then resolves the issue.โ
The Results
Comprehensive cloud support from complementary cloud partners
Working with both DoiT and BigData Boutique in tandem has given Tastewise comprehensive cloud expertise to help grow its business.ย
With BigData Boutique, it has the Elasticsearch expertise to provide faster, smarter insights to customers, with no more timeouts. With DoiT, Tastewise can scale confidently with predictable unit economics and deliver more for less.
Two full-time engineer salaries saved
Relying on DoiT and BigData Boutique for their expertise means Tastewise reclaims capacity equivalent to two full-time engineers. โIn simple terms, BigData Boutique saves us the cost of a dedicated Elasticsearch engineer and DoiT saves us the cost of a FinOps engineer,โ Gill explains.ย
Keeping its team lean in this way allows the company to invest its resources in developing core products and services.
40% savings on Elasticsearch maintenance with BigData Boutiqueโs expertise
By running Elasticsearch on its own Amazon EC2 instances and relying on BigData Boutique to help maintain the service, Tastewise reduced operating costs by ~40% versus a managed service while retaining control.
$1,000 a month saved on ElastiCache instances
DoiTโs recommendation to switch ElastiCache from on-demand instances to reserved instances saved Tastewise $1,000 a month.
$1,000s saved with Anomaly Detection
Being able to spot and fix cost anomalies quickly with Anomaly Detection has saved Tastewise thousands of dollars on its monthly cloud bill.
36% saved on EC2 workloads with Flexsave for Compute
With Flexsave for Computeโข identifying and applying discounted rates to Tastewiseโs EC2 workloads, Tastewise saves 5-10% a month on those workloads, amounting to an all-time effective savings rate of 36%.
What's Next?
With DoiT helping to manage its cloud costs, and Pulse giving it visibility of its Elasticsearch clusters, Tastewise can now focus on scaling up its business, confident that any issues will be quickly identified and remedied by its cloud partners.ย
As Gill explains, โIf we want to use a new service, we can simply go ahead, safe in the knowledge that DoiT will help us to save money as we use it. And with BigDataBoutique, we can continue to use Pulse, as well as the companyโs Elasticsearch expertise, to keep optimizing our search capabilities. I can sleep soundly at night knowing that with DoiT and BigData Boutique, we have the expertise to handle any challenges that might arise.โ