Hi [[ session.user.profile.firstName ]]

High-Load Storage of Users’ Actions with Scylla and HD

The presentation gives a brief overview of the high-load service that stores users' actions using HDDs. The given service is able to serve up to 240k writes per second in less than 2ms 95 percentile with just a few ScyllaDB nodes packed with HDDs. Hardware setup, specification for cluster HDD, live load numbers and latencies achieved are given. The problems we encountered with HDD setup are described along with the possible solutions to them.
Recorded May 22 2021 12 mins
Your place is confirmed,
we'll send you email reminders
Presented by
Kirill Alekseev, Software Engineering Team Lead, Mail.Ru
Presentation preview: High-Load Storage of Users’ Actions with Scylla and HD

Network with like-minded attendees

  • [[ session.user.profile.displayName ]]
    Add a photo
    • [[ session.user.profile.displayName ]]
    • [[ session.user.profile.jobTitle ]]
    • [[ session.user.profile.companyName ]]
    • [[ userProfileTemplateHelper.getLocation(session.user.profile) ]]
  • [[ card.displayName ]]
    • [[ card.displayName ]]
    • [[ card.jobTitle ]]
    • [[ card.companyName ]]
    • [[ userProfileTemplateHelper.getLocation(card) ]]
  • Channel
  • Channel profile
  • Disney+ Hotstar: Scaling NoSQL for Millions of Video On-Demand Users Recorded: Jul 7 2021 52 mins
    Vamsi Subhash Achanta, Architect, Disney+ Hotstar, Balakrishnan Kaliyamoorthy, Senior Data Engineer, Disney+ Hotstar
    Disney+ Hotstar is the fastest growing branch of Disney+. Join Disney+ Hotstar Architect Vamsi Subhash and senior data engineer Balakrishnan Kaliyamoorthy to learn:

    - How Disney+ Hotstar architected their systems to handle massive data loads
    - Why they chose to replace both Redis and Elasticsearch
    - Their requirements for massively scalable data infrastructure and evolving data models
    - How they migrated their data to Scylla Cloud, ScyllaDB’s fully managed NoSQL database-as-a-service, without suffering downtime
  • NoSQL and NewSQL: Tradeoffs Between Scalable Performance & Consistency Recorded: Jun 3 2021 58 mins
    Avishai Ish-Shalom, Developer Advocate at ScyllaDB
    Traditional relational databases provide strong consistency, but they can be difficult to scale. Non-relational databases scale more easily, but generally offer less strict consistency guarantees. Now a new class of database, known as NewSQL, promises both strong consistency and the ability to scale.

    In this webinar, ScyllaDB developer evangelist Avishai Ish Shalom will compare NoSQL and NewSQL databases. We will look at the significant architectural differences between the two; tradeoffs between availability, scalable performance and consistency; data models and other considerations.

    We will also share benchmark results to display the performance implications of NoSQL versus NewSQL. For this purpose we will compare Scylla with best-in-class NewSQL database CockroachDB.

    By attending this webinar you’ll get:
    - An overview of NoSQL and NewSQL
    - Highlight of architectural differences between NoSQL and NewSQL
    - Tradeoffs between scalable performance and consistency in NoSQL and NewSQL
    - YCSB Benchmark results for Scylla and CockroachDB
  • Fast NoSQL on HDDs? Recorded: May 24 2021 52 mins
    Kirill Alekseev, Software Engineering Team Lead, Mail.Ru
    Learn the secrets behind how Russia’s #1 internet provider achieves high database performance at low cost.

    What if you could store huge amounts of data using traditional low-cost Hard Disk Drives (HDDs) while still maintaining low single-digit millisecond latencies? That’s just what they did at Mail.Ru, the largest email service provider in Russia.

    Join Kirill Alekseev, software engineering team lead at Mail.Ru, for an overview of the extremely efficient approach his team took when implementing a high-load NoSQL service for storing user actions. You’ll see how Mail.Ru uses Scylla to serve up to 240k writes per second with 95 percentile latencies of less than 2ms – using a Scylla cluster running on HDDs.

    From attending this session you’ll learn:

    - How to build a large time-series storage with inexpensive HDDs while maintaining consistent low latencies under heavy write load, including the hardware and cluster setup
    - Problems they encountered with their HDD setup and how they solved them
    - Overview of the service and the logic it implies: how they handled retries and sped up API response with concurrent reads
    - Tips & tricks they picked up along the way
  • Torture Testing NoSQL: Putting Scylla to the Jepsen Test Recorded: May 23 2021 59 mins
    Kyle Kingsbury, creator of Jepsen
    Kyle Kingsbury (aka ‘Aphyr’) tortures databases for a living. He pokes and prods and stresses them with his Jepsen test to look for safety violations across distributed systems. Kyle worked with the ScyllaDB team to identify multiple issues in Scylla's lightweight transactions (LWT) implementation, including split-brain behavior and write loss in healthy clusters.

    Tune in to learn about distributed database testing techniques, Scylla’s concurrency properties, Kyle’s deep dive into transaction analysis and how to use Scylla safely. Kyle will talk about the specific issues he found in Scylla, and how the ScyllaDB team addressed them.
  • Zillow: Optimistic Concurrency Using Write-time Timestamps Recorded: May 22 2021 11 mins
    Dan Podhola, Principal Software Engineer, Zillow
    Consuming messages from multiple sources in a scalable service sharing the same Scylla cluster can be problematic, especially when those sources provide data from two queues each and cannot deliver guaranteed in-order messages. At Zillow, we provide the Scylla write timestamp and employ a couple other tricks at Zillow to provide correct and consistent data to our consuming services and avoid doing transactions. Additionally, the Zillow data model is a simple compound key that contains a large binary Avro document.
  • Ticketmaster Performance Test: Scylla, DataStax DSE and Apache Cassandra Recorded: May 22 2021 15 mins
    Linda Xu, Vice President of Data Platform, Ticketmaster
    Ticketmaster is part of Live Nation Entertainment, the world's leading live entertainment company. Learn why they went with Scylla after conducting performance testing between Scylla, Apache Cassandra and DataStax Enterprise.
  • The All-New Scylla Monitoring Stack Recorded: May 22 2021 10 mins
    Amnon Heiman, Software Developer, ScyllaDB
    Scylla Monitoring has been enhanced to be a one stop shop to monitor Scylla. To achieve this, scylla-monitoring now includes a CQL interface and log collection analysis. The combination of matrices, alerts, logs & CQL allows not only a single point to view all the information but also to advise you what may be the source of a performance/latency degradation (malfunctioning node.)
  • Streaming Data from Scylla to Kafka Recorded: May 22 2021 12 mins
    Tim Berglund, Senior Director of Developer Advocacy, Confluent
    Change Data Capture (CDC) and the Kafka Scylla Connector now allow Scylla to act as a data producer for Kafka streams. Discover how combining Scylla with the Confluent platform allows to maximize the value of NoSQL data by introducing Scylla as a key component of event driven architecture and enables streaming database changes, enriching these updates via message transformations, and allowing users to efficiently run data pipelines.
  • Scylla’s Journey Towards Being an Elastic Cloud Native Database Recorded: May 22 2021 9 mins
    Shlomi Livne, VP of R&D, ScyllaDB
    Cloud Native Databases are required to scale while serving the increase in online workload with a minimal disruption and complete it as fast as possible. In this session we will review the different components that are stressed in scaling scenarios and present work we have done over the year to improve Scylla’s elasticity as we enhance it to be a true Cloud Native Database.
  • Scylla Performance Enhancements: Year in Review Recorded: May 22 2021 12 mins
    Tomasz Grabiec, Distinguished Engineer, ScyllaDB
    The talk covers most of the performance enhancements introduced to Scylla over the past 12 months. As the throughput was very good before, we focused on Scylla’s behavior under all types of workloads and data models. Scylla improved its latency under all scenarios, improving the behaviors of data models such as large partitions and time series, improvement of the I/O scheduler and behavior of streaming and repair.
  • The True Cost of NoSQL DBaaS Options Recorded: May 22 2021 47 mins
    Eyal Gutkind, VP of Solutions, ScyllaDB
    Many NoSQL DBaaS vendors limit what cloud platform you can run on, the size of the data you can run and require you to over-provision cloud infrastructure resources while failing to deliver performance and low latency at scale.

    In this session, we will compare the performance and Total Cost of Ownership (TCO) of competing NoSQL DBaaS offerings. We will also review how to migrate to Scylla Cloud, our fully managed database service.

    You will learn:

    - The true cost of ownership for selected NoSQL DBaaS offerings
    - The 8 essentials for selecting a NoSQL DBaaS
    - Migration options from Apache Cassandra, DynamoDB and other databases
  • The Magic of Scylla Operator 1.0 Powered by Kubernetes Recorded: May 22 2021 13 mins
    Maciej Zimnoch, Software Engineer, ScyllaDB
    Scylla Kubernetes support is long anticipated. Powered by Kubernetes, Scylla becomes easy to manage. Scale out, rolling upgrade, automatic configuration change in canary deployment style and so forth. The talk will demonstrate Scylla Operator 1.0 and go over its supported feature set and describe the roadmap ahead.
  • Scylla Cloud on Display Recorded: May 22 2021 12 mins
    Erez Mazor, Engineering Manager, ScyllaDB
    Join us for a session showcasing Scylla Cloud and how easy it is to integrate and connect with your applications and forget about all aspects of managing, securing, scaling and maintaining a distributed, always-on database. We will demo Scylla Cloud and present deployment and account options over various cloud providers.
  • Scylla and Jepsen Tests Recorded: May 22 2021 20 mins
    Kyle Kingsbury, Principal, Jepsen
    The Jepsen framework is a tool for testing databases, combining fault injection, generative testing, and sophisticated safety checkers to find concurrency bugs and availability issues. We worked with the ScyllaDB team to identify multiple safety issues in Scylla's lightweight transactions (LWT) layer, including split-brain behavior and write loss in healthy clusters. In this talk, we'll discuss testing techniques more generally, and what we found (and what was fixed) in ScyllaDB.
  • Scylla @ GumGum: Contextual Intelligence Ads Recorded: May 22 2021 15 mins
    Keith Sader, Engineering Manager of the Ad Server (Demand) team, GumGum
    Gumgum is a global media company specializing in contextual intelligence. Find out how and why they moved to Scylla Cloud to meet their growing customer demands and deliver on their promise of future proofed solutions.
  • Scylla @ Disney+ Hotstar: On-Demand Streaming Service Recorded: May 22 2021 15 mins
    Vamsi Subhash Achanta, Architect, Disney+ Hotstar, Balakrishnan Kaliyamoorthy, Senior Data Engineer, Disney+ Hotstar
    Disney+ Hotstar provides on-demand streaming to more than 18.5 million paid subscribers and 300 million monthly active users. Hotstar's India Premier League (IPL) coverage in 2020 made it the most widely streamed sports events in the world to date. Learn why they chose Scylla Cloud to replace both Redis and Elasticsearch, and how Hotstar data usage was migrated with zero downtime.
  • Raft in Scylla Recorded: May 22 2021 15 mins
    Konstantin "Kostja" Osipov, Software Team Lead, ScyllaDB
    Agreement in a distributed system is complicated but required. Scylla gained lightweight transactions through Paxos but the latter has a cost of 3X roundtrips. Raft can allow consistent transactions without the performance penalty. Beyond LWT, we plan to integrate Raft with most aspects of Scylla making a leap forward in manageability and consistency.
  • Prepare to Love Scylla Even More Recorded: May 22 2021 19 mins
    Dor Laor, CEO and Co-Founder, ScyllaDB
    The ScyllaDB team has accomplished a lot since our last Scylla Summit. For example, our DynamoDB API is now GA, running on any cloud or on-prem, and certified as AWS Outposts ready. Our team is busy working on even bigger advancements to widen the gap between Scylla and other NoSQL databases. Join ScyllaDB CEO and co-founder Dor Laor to hear about these new initiatives and to see what the future of Scylla holds.
  • Getting the Scylla Shard-Aware Drivers Faster Recorded: May 22 2021 10 mins
    Alexys Jacob, CTO, Numberly
    Alexys explains how Scylla shard-aware drivers are implemented and why Scylla benefits from them. Taking the Python shard-aware driver as an example, Alexys will demonstrate how recent shard-aware drivers can leverage the new shard allocation algorithm that Scylla 4.3 brings to the table, and how to make use of it from a developer or administrator point of view. Alexys will showcase benefits from real production graphs observed at Numberly.
  • Meeting the Challenge of Big Data and NoSQL Databases Recorded: May 22 2021 29 mins
    Avi Kivity, CEO & Co-Founder, ScyllaDB
    Scylla has emerged as the most cost-effective way to serve large amounts of data. In this keynote we will cover what makes Scylla such a great match for big data applications, what we’ve done over the past year, and our plans for the near future: performance, reliability, manageability, and above all, making Scylla easy to consume in the modern cloud environment.
The Real-Time Big Data Database
The Scylla NoSQL database embraces a shared-nothing approach that increases throughput and storage capacity as much as 10X. It comes in open source, enterprise and database-as-a-service options. Comcast, Discord, Grab, Medium, Starbucks, Ola Cabs, Samsung, IBM, Investing.com, Zillow and many more leading companies have adopted Scylla to realize order-of-magnitude performance improvements and reduce hardware costs. For more information: ScyllaDB.com.

Embed in website or blog

Successfully added emails: 0
Remove all
  • Title: High-Load Storage of Users’ Actions with Scylla and HD
  • Live at: May 22 2021 10:30 am
  • Presented by: Kirill Alekseev, Software Engineering Team Lead, Mail.Ru
  • From:
Your email has been sent.
or close