Loading…
November 18, 2019 | San Diego, California
View More Details & Register Here

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

Monday, November 18
 

8:00am PST

9:00am PST

Orientation and Community Update - Alex Miller, Apple
Speakers
AM

Alex Miller

Software Engineer, Apple
Alex Miller has been working on FoundationDB for 3 years. He recently changed the transaction logs in FoundationDB to be significantly more performant during region failures.


Monday November 18, 2019 9:00am - 9:20am PST
Grand Ballroom 5

9:20am PST

Retrofitting a Foundation for CouchDB - Adam Kocoloski, IBM
FoundationDB tackles some wickedly hard problems in distributed systems and offers developers a reliable toolkit for building modern data services. The resulting layered approach to database development is an attractive architecture for “new construction”, but how well does it work in a renovation?

The Apache CouchDB project is adopting FoundationDB as its core persistence layer. CouchDB is a JSON document store that implements many standard DB features (primary keys, indexes, materialized views) as well as some more exotic ones, including active/active replication across any user-defined set of databases. We’ll discuss CouchDB’s rationale for adopting FoundationDB and report our progress on this journey. We’ll provide comparisons to extant layers like RecordLayer and DocumentLayer, and share some of our experiences running an open source database project over the past decade.

Speakers
avatar for Adam Kocoloski

Adam Kocoloski

IBM Fellow, VP & CTO, Cloud Databases, IBM
Adam is an IBM Fellow, VP and CTO, Cloud Databases. He joined IBM in 2014 via the acquisition of Cloudant, where he built a highly available, scalable database and drove the development of the systems required to offer the database as a cloud service. Adam has a long track record... Read More →



Monday November 18, 2019 9:20am - 9:50am PST
Grand Ballroom 5
  General Session
  • Session Slides Included Yes

9:50am PST

FoundationDB Roadmap: Past and Future - Evan Tschannen, Apple
During the last Summit, a number of ongoing projects were presented.  This will be an update on their status and other notable features that have occurred in the past year, as well as an overview of what work is currently ongoing that will be released over the next year.

Speakers
ET

Evan Tschannen

Software Engineer, Apple
Evan Tschannen has been working on FoundationDB for over 8 years. He has implemented many of the critical components of the system. He now manages the development team for FoundationDB at Apple.



Monday November 18, 2019 9:50am - 10:10am PST
Grand Ballroom 5
  General Session
  • Session Slides Included Yes

10:10am PST

Break
Monday November 18, 2019 10:10am - 10:40am PST
Grand Ballroom 5 Foyer

10:40am PST

Time Series and FoundationDB: Millions of Writes/s and 10x Compression in 2000 Lines of Go - Richard Artoul, Uber
Richard spent the last two years developing a time series database called M3DB that was designed from the ground up (including a custom storage engine) to address the issue of ingesting and compressing million of time series values per second.

In this talk, Richard will revisit the original problem that M3DB was designed for, but from a new perspective. Could M3DB have been built on top of FoundationDB instead?

Over the course of the talk, Richard will walk participants through the process of designing a distributed system on top of FoundationDB that ingests millions of data points per second and achieves industry-standard levels of compression, all in less than 2,000 lines of Go.

The talk will include detailed implementation details such as code samples, architecture diagrams, and an overview of how the FoundationDB key/values were structured.

Speakers
RA

Richard Artoul

Software Engineer, Uber
Richard Artoul is a software engineer at Uber and tech lead on the M3DB project, Uber's open-source distributed time series database.



Monday November 18, 2019 10:40am - 11:00am PST
Grand Ballroom 5
  Case Studies
  • Session Slides Included Yes

10:40am PST

Flow: An Introduction - Markus Pilman, Snowflake
Flow is the programming language FoundationDB is implemented in. It brings an actor-based concurrency model to C++. Flow is specifically optimized two write I/O heavy, highly concurrent software.

This talk is meant for people who either are interested in contributing code to core FoundationDB, want to be able to read FoundationDB’s source code, are interested in distributed systems and want to evaluate whether Flow could be a good choice for them, or just want to learn a cool new language. You will learn how to write simple actors, how the underlying code works, and what the most common pitfalls are people encounter when writing in Flow.

Speakers
avatar for Markus Pilman

Markus Pilman

Software Engineer, Snowflake
Markus Pilman is the engineering lead of the FoundationDB team at Snowflake Computing. He joined Snowflake two years ago after graduating as a Doctor of Sciences from ETH Zurich.He was the author or co-author of and presenter of several papers in the academic database community.



Monday November 18, 2019 10:40am - 11:00am PST
Grand Ballroom 8
  FDB Internals
  • Session Slides Included Yes

11:00am PST

Evaluation of FoundationDB in Financial Services - Alok Madhukar, Rishabh Maheshwari & Subramaniam Ramamoorthi, Goldman Sachs
Within Goldman Sachs we have an in-house data database for trading and risk management called SecDB. SecDB is the single system that captures all risk artifacts for the firm and is the backbone for pricing and risk.

SecDB is a distributed in-memory blob-store with multiple write-masters that spread across regions. There is a custom synchronization protocol to replay transactions between different write masters. The system is eventually consistent and the trade model is designed to be conflict-free in most of the cases. SecDB was built more than 25 years back when nothing like it existed in the market. As the industry landscape is changing, we need to re-design our systems to be future-ready As we evaluated multiple options, we found FoundationDB to be a good match for many of our use cases. In our talk and paper, we’ll discuss good, bad and ugly of FoundationDB for our use-cases.

Speakers
AM

Alok Madhukar

Technology Fellow, Goldman Sachs
Alok Madhukar is a Technology Fellow (Vice President) in Core Engineering Division of Goldman Sachs where he is responsible for architecture for firm’s strategic trading and risk calculation platform.  Alok joined the firm in 2006.  Prior to joining the firm Alok held various... Read More →
SR

Subramaniam Ramamoorthi

Vice President, Goldman Sachs
Subramaniam Ramamoorthi is a Vice President in Goldman Sachs and looks after the design of firm's strategic trading platform.
RM

Rishabh Maheshwari

Vice President, Goldman Sachs
Rishabh Maheshwari is a Vice President in Core Engineering at Goldman Sachs and looks after various SecDb Architecture components. He holds a Master’s Degree in Electrical Engineering from IIT Kanpur. Rishabh is passionate about system design and is a database enthusiast. He enjoys... Read More →



Monday November 18, 2019 11:00am - 11:20am PST
Grand Ballroom 5
  Case Studies
  • Session Slides Included Yes

11:00am PST

Performance Profiling in FoundationDB - Diego Didona, IBM Research Zurich
Understanding the performance of a system is paramount to identify which components and functions need to be improved to obtain better performance.  In this talk, we will present how we use the Linux performance analysis tools, e.g., perf, to analyze the performance of FoundationDB. Specifically, we will describe how we use these tools to identify performance bottlenecks and areas for potential performance improvements within FoundationDB. We will use concrete examples taken from the analysis of both the client library and components of the server, e.g., the new Redwood storage engine.

Speakers
DD

Diego Didona

Visiting Scientist, IBM Research Zürich
Diego Didona is a Visiting Scientist at the Cloud and Computing Infrastructure at IBM Research Zürich. His research interests include storage systems, both persistent and in-memory, and data consistency in geo-replicated settings.



Monday November 18, 2019 11:00am - 11:20am PST
Grand Ballroom 8
  FDB Internals
  • Session Slides Included Yes

11:20am PST

VMware Network Insight's Configuration-Store Layer - Gaurav Agarwal, VMware
This talk introduces the heterogeneous infrastructure stack in an enterprise product and the motivation towards consolidating it into FDB - starting from a PostgreSQL based configuration time-series store.

It will outline our journey as we progressed from evaluating FDB’s concepts and guarantees, developing frameworks and practices for modeling relational tables as KV-ranges, to using advanced FDB concepts like VersionStamp, Conflicts, GRV-Cache, Locality, to build tighter and performant models; and challenges such as higher latencies, transaction/row size and time limits, and their impact on design choices.

We will describe our tooling and integrations to get visibility into (a) application layer’s interactions with FDB, (b) areas for optimization (c) health of FDB cluster. We will discuss the challenges when troubleshooting FDB, and possible enhancements that could simplify it.

Speakers
GA

Gaurav Agarwal

Staff-2 Engineer, VMware, Inc.
I work on VMWare's Network Insight's infrastructure layer team to build robust and scalable distributed storage systems: configuration-stores, metric time-series stores and flow analysis systems. In the past, I have had an opportunity to work on very large scale data processing and... Read More →



Monday November 18, 2019 11:20am - 11:40am PST
Grand Ballroom 5
  Case Studies
  • Session Slides Included Yes

11:20am PST

A New Radix-Tree Based Memory Storage Engine - Mengran Wang, VMware
This talk describes the development of a new FDB memory storage engine based on the radix tree data structure. The default memory storage engine is a balanced binary search tree. Because Wavefront’s use case involves a large amount of prefix-overlap in keys, we decided to adopt a different data structure, the radix tree, to make our FDB clusters more space-efficient. This trie-based KV store has two main advantages:
(1) It is more space-efficient than the default engine when there is a large amount of key overlap (as in our time series workloads).
(2) The height and complexity of a radix tree depend on the length of the keys, not on the number of elements

Our presentation will include a brief introduction to radix tries followed by a deep dive into some of the optimizations we needed to make. Finally, we will show the performance comparison results using real-life Wavefront workloads.

Speakers
MW

Mengran Wang

Member of Technical Staff, VMware
Mengran Wang is a software engineer at VMware, Wavefront team, a real-time metrics monitoring and streaming analytics platform. She’s working on Foundation DB related projects, primarily focus on memory storage engine optimization/development.  



Monday November 18, 2019 11:20am - 11:40am PST
Grand Ballroom 8
  FDB Internals
  • Session Slides Included Yes

11:40am PST

Using FoundationDB and the FDB Record Layer to Build CloudKit - Scott Gray, Apple
In this case study, we will highlight some key features of FoundationDB and the Record Layer that help us build CloudKit, Apple’s cloud storage system for structured data. We will focus on three key areas where FoundationDB and the Record Layer have unlocked large benefits for CloudKit. First, FoundationDB’s arbitrary multi-key ACID transactions have allowed us to implement advanced secondary indexing at scale, including our recently-developed transactional full-text search system. Second, the Record Layer’s version indexes have greatly increased the scalability of CloudKit’s core cross-device synchronization functionality. Finally, FoundationDB’s best-in-class reliability has simplified our operations, especially while recovering from failures. All together, FoundationDB has allowed us to provide richer APIs, greater scalability, and higher availability across billions of databases.

Speakers
avatar for Scott Gray

Scott Gray

Software Engineer, Apple
Scott is an engineer at Apple working on CloudKit, Apple's cloud database for structured data.



Monday November 18, 2019 11:40am - 12:00pm PST
Grand Ballroom 5
  Case Studies
  • Session Slides Included Yes

11:40am PST

Redwood Storage Engine Update - Steve Atherton, Apple
Redwood is FoundationDB’s new B-Tree based storage engine which is aimed at offering higher I/O throughput and smaller on-disk footprints. This talk will give an overview and update on the project, covering current features and caveats to consider, performance characteristics for different workloads, and planned future optimizations.

Speakers
SA

Steve Atherton

Software Engineer, Apple
Steve Atherton is a core developer on FoundationDB at Apple.



Monday November 18, 2019 11:40am - 12:00pm PST
Grand Ballroom 8
  FDB Internals
  • Session Slides Included Yes

12:00pm PST

Lunch
Lunch will be provided.

Monday November 18, 2019 12:00pm - 1:00pm PST
Grand Ballroom 5

1:00pm PST

Consistent Caching in FoundationDB - Xin Dong, Apple & Neelam Goyal, Snowflake
We’re presenting a new FoundationDB feature that is currently worked on as a collaboration project between Apple and Snowflake. If a key-range (or a single key) is very read-hot, the storage processes will start to run out of resources. Currently, the only mitigation for this is to either change the access pattern or implement a caching functionality on top of FoundationDB. Both of these are difficult to implement and often need special implementations depending on the consistency requirement. We’re currently working on a solution for hot read ranges that attacks this problem from two angles: (1) allow for some key-ranges to be held in memory and (2) increase the replication factor for read-hot ranges. (1) will take rad-load off disks while (2) will allow using more CPUs to handle the load. This cache will provide the same consistency guarantees as the rest of FoundationDB. The administrator will have the option to define ranges and replication factors for certain ranges and additionally, FoundationDB will detect ranges that are read-hot and not write-hot and will start to cache them automatically


Speakers
NG

Neelam Goyal

Software Engineer, Snowflake
XD

Xin Dong

Software Engineer, Apple



Monday November 18, 2019 1:00pm - 1:12pm PST
Grand Ballroom 5
  Lightning Talk
  • Session Slides Included Yes

1:12pm PST

Meta-Data Caching and Record Store Scalability - Alec Grieser, Apple
One common pattern for applications and layers on top of FDB is to store meta-data about the current schema in the database so that schema changes can be applied transactionally with data updates. But while this is useful and sometimes vital for ensuring that the data and schema do not drift out of sync, constant reads to a single meta-data key can be a performance bottleneck that can limit the total scale of a single data store. Using features included in FoundationDB 6.1, the Record Layer was able to add a caching mechanism for schema meta-data that still provided fully transactional updates and allowed us to push past that limit.

Speakers
avatar for Alec Grieser

Alec Grieser

Software Engineer, Apple
Alec is a software engineer at Apple working on CloudKit, a managed cloud database. He is one of the contributors to the FoundationDB Record Layer, an open source library which provides roughly relational semantics on top of the FoundationDB key-value store. He has spoken about FoundationDB... Read More →



Monday November 18, 2019 1:12pm - 1:19pm PST
Grand Ballroom 5
  Lightning Talk
  • Session Slides Included Yes

1:19pm PST

Unlucky Simulation - Andrew Noyes, Snowflake Computing
Foundationdb's correctness testing involves using a fair PRNG in a simulation to make decisions that the application cannot control such as when messages get delivered, when machines reboot, and when processes crash. The idea is that with a sufficiently large number of simulations we achieve adequate code coverage. However, most of these simulations do not actually increase code coverage.

In this talk, we evaluate the use of a coverage-guided fuzzer as a "random" number generator for simulation and compare the compute resources required for catching rare bugs.

Speakers
AN

Andrew Noyes

Software Engineer, Snowflake Computing
Andrew is a software engineer working at Snowflake Computing on FoundationDB development.



Monday November 18, 2019 1:19pm - 1:26pm PST
Grand Ballroom 5
  Lightning Talk
  • Session Slides Included Yes

1:26pm PST

New Snowflake Capabilities and What They Mean for FoundationDB - Torsten Grabs, Snowflake
Snowflake as a cloud-built data warehouse heavily relies on FoundationDB for metadata management. In this talk, we will explore some of Snowflake's newest capabilities such as database replication, data sharing, data pipelines and data lake integration. Their implementation presents us with interesting challenges, both for metadata management in FoundationDB and for elastic and scalable cloud data processing in general. We will discuss how we have tackled these challenges in Snowflake and where we see additional work in FoundationDB and beyond.

Speakers
avatar for Torsten Grabs

Torsten Grabs

Product Manager, Snowflake
Torsten currently serves as Director of Product Management for Snowflake where he heads product management for Snowflake's data pipeline, data lake and developer strategy.  Before joining Snowflake, Torsten spent more than a decade at Microsoft in the SQL Server product group in... Read More →


Monday November 18, 2019 1:26pm - 1:33pm PST
Grand Ballroom 5

1:33pm PST

Scale & Security for Data Workloads on IBM Cloud - Jason McGee, IBM
Every containerized app needs scalable and secure data. In this talk, you’ll hear about our container journey. When IBM began moving workloads to Kubernetes, scale and security were our foremost necessities. Today, we manage 16,000+ clusters on our public cloud. And those clusters are driving some immense workloads, such as The Weather Company's 25 billion forecasts a day. Kubernetes clusters provide multiple tools to handle elasticity: worker pools, pod autoscaling, and Knative serverless compute, to name a few.  But scale is only part of our story. Financial Services and Healthcare companies also trust IBM Cloud with sensitive persistent data. These companies leverage our automated expertise of cloud services for patching, vulnerability detection, network isolation, and encryption, so they're free to focus on apps.  With years of running enterprise workloads in our public cloud, open-source tools like FoundationDB and OpenShift are the kind of secure and scalable technologies our users count on.

Speakers
avatar for Jason McGee

Jason McGee

IBM Fellow, VP and CTO, IBM Cloud Platform, IBM
Jason is currently responsible for the IBM Cloud’s platform services, including Kubernetes, Functions, Cloud Foundry, Kafka event streams, Logging, Monitoring, Container Registry, Schematics, Terraform and Activity Tracker. Jason is also responsible for the technical strategy and... Read More →



Monday November 18, 2019 1:33pm - 1:40pm PST
Grand Ballroom 5
  Lightning Talk
  • Session Slides Included Yes

1:40pm PST

FoundationDB as Datastore for Training Deep Learning Models - Kannan Dorairaj, Axstrm
Axstrm is a data services and solutions company focused on building data stores for computer vision, deep learning, artificial intelligence, and analytics solutions. We have been using FoundationDB as our go-to database for our custom data stores. We will be going over a real-life scenario to build self-trained micro models from dynamically distributed data coming in from different sources and trained to do a specific objective. In the process of taking FoundationDB to mainstream we have developed a structured methodology to monitor and maintain FoundationDB in production environments.

Speakers
KD

Kannan Dorairaj

CEO & Chief Architect, Axstrm
An entrepreneurial thought leader with 20+ years of experience in enterprise software architecture, database and technical leadership.



Monday November 18, 2019 1:40pm - 1:47pm PST
Grand Ballroom 5
  Lightning Talk
  • Session Slides Included Yes

1:50pm PST

Rearchitecting CouchDB Secondary Indexes on Top of FoundationDB - Garren Smith, Red Comet Labs
How do you rewrite a database that’s been around since 2005, so your users never notice a change in behavior or have to sacrifice features, but get all the improvements in performance and consistency that FoundationDB brings?

That’s the challenge we’ve taken on in rewriting Apache CouchDB to run on top of FDB. One of the biggest architectural redesigns is CouchDB’s secondary indexes: Map/Reduce, Mango, and Text Search.

This talk covers the architectural solutions we’ve found for challenges such as: finding the most suitable data model for each index; managing global index state and avoiding conflicts when building indexes; ensuring that the results returned are sorted correctly; and using Apache Lucene on top of FDB for Text Search.

I’ll also share the results of the benchmarking we’ve done, along with further plans we have to improve the architecture of each index.

Speakers
avatar for Garren Smith

Garren Smith

CouchDB Developer, Red Comet Labs
Garren is a member of the Apache CouchDB Project Management Committee and a JavaScript and Erlang developer, who enjoys working on distributed systems. He is currently working on the IBM Cloudant Database, which powers IBM Cloud and has CouchDB at its core. He is part of the team... Read More →



Monday November 18, 2019 1:50pm - 2:10pm PST
Grand Ballroom 8
  Building Layers
  • Session Slides Included Yes

1:50pm PST

FoundationDB and Node.js: Year in Production - Steve Korshakov, Openland
About a year ago we moved all our stack to FoundationDB, in the last year we have built a multi-functional library (http://fdb.openland.com) to work with FDB in NodeJS that can: Distributed Locks, Work Queue, Entity Layer (very similar to Record Layer), Distributed Random ID generator, Singleton Workers and so on. I want to share my experience using FDB in production for building a messaging app and sharing this library to the community.

Speakers
SK

Steve Korshakov

CTO, Openland
Founder/CTO of Openland, messaging for work, worked at Telegram as SSWE.


FDB pdf

Monday November 18, 2019 1:50pm - 2:10pm PST
Grand Ballroom 5
  Real World FDB
  • Session Slides Included Yes

2:10pm PST

FoundationDB as a Storage Engine for Irmin - Andreas Garnæs, Peakon
Irmin is an open-source distributed database built on the same principles as Git. Like Git, Irmin provides snapshotting, branching and merging operations over storage and can communicate via Git both on-disk and remotely.

Further, Irmin is highly configurable and offers pluggable storage engines. FoundationDB has proven to be a very attractive option in this regard, due to its reliability, strong atomicity guarantees and rich feature set, such as watches. This has resulted in `irmin-fdb`, a storage engine for Irmin based on FoundationDB. `irmin-fdb` can be perceived as a layer, that allows you to interact with FoundationDB in yet another mode.

Irmin is implemented in OCaml, a strongly typed functional language. Developing `irmin-fdb` thus required developing a FoundationCB client for OCaml.

Speakers
avatar for Andreas Garnæs

Andreas Garnæs

Engineering Manager, Peakon
Andreas is an Engineering Manager at Peakon, who are building the world's leading platform for measuring and improving Employee Engagement. Andreas is an active contributor to Irmin, an open source distributed database built on the same principles as Git. In particular, Andreas has... Read More →



Monday November 18, 2019 2:10pm - 2:30pm PST
Grand Ballroom 8
  Building Layers
  • Session Slides Included Yes

2:10pm PST

Restoring FDB after Catastrophic Accidents - Brandon Burton & Matthew Zeier, Wavefront
How do you backup (and restore) a multi-petabyte FoundationDB cluster? How do you do it when you've suffered a catastrophic event?

We'll cover three real-world examples of failures and how we recovered from them, including tips we wish we knew. Some will involve FDBDR and others will involve Linux filesystem snapshots and as an added bonus some tricks too, in near real-time, duplicate a running production cluster elsewhere.

Speakers
avatar for Brandon Burton

Brandon Burton

Cloud Operations, Wavefront by VMware
avatar for Matthew Zeier

Matthew Zeier

Sr. Manager, Wavefront Operations, Wavefront by VMware, Wavefront
Founder, #BeachOps. Sr Manager. Runs Wavefront Operations. Matthew believes the only thing that matters is Customer Happiness and brings this as the guiding principle in Operations. He founded #BeachOps because I'd rather build for lazy than build for busy. And going to the beach... Read More →



Monday November 18, 2019 2:10pm - 2:30pm PST
Grand Ballroom 5
  Real World FDB
  • Session Slides Included Yes

2:30pm PST

Building a Performant Configuration-Store Layer - Guarav Agarwal, VMware
Frequently, one needs an “ordered-log” model that has the characteristic of appending to the End of the key-space. For example, one such use-case is to model a “change-log” table, that maintains an ordered-history of changes that are happening to some other data/state. Typically, in FDB, this model is implemented by prefixing the change-keys with VersionStamp, to provide a strong ordering between change records.

However, this creates write-hotspots due to the inherent concentration of new rows to a specific shard at any given time.

We would like to present a technique that can help solve this problem by avoiding this severe concentration of keys, while still preserving the desired ordering, at the cost of some overhead at read-time, and some conflicts at write time. The conflicts could possibly be solved with potential enhancement in FDB core (https://bit.ly/2M6twot).

Speakers
GA

Gaurav Agarwal

Staff-2 Engineer, VMware, Inc.
I work on VMWare's Network Insight's infrastructure layer team to build robust and scalable distributed storage systems: configuration-stores, metric time-series stores and flow analysis systems. In the past, I have had an opportunity to work on very large scale data processing and... Read More →



Monday November 18, 2019 2:30pm - 2:50pm PST
Grand Ballroom 8
  Building Layers
  • Session Slides Included Yes

2:30pm PST

Managing FoundationDB at Scale - John Brownlee, Apple
Apple runs hundreds of FoundationDB clusters with high availability requirements. Running FoundationDB at this scale requires careful design and specialized tooling. In this presentation, we will explore the core workflows we use to manage cluster lifecycle, the high-level design considerations in managing FoundationDB, and how we handle common operational problems.

Speakers
JB

John Brownlee

Software Engineer, Apple
I'm a developer on the FoundationDB SRE team at Apple.



Monday November 18, 2019 2:30pm - 2:50pm PST
Grand Ballroom 5
  Real World FDB
  • Session Slides Included Yes

2:50pm PST

A ZooKeeper Layer for FoundationDB - Paul Hemberger, HubSpot
ZooKeeper plays a central role in coordinating many distributed systems. It is also itself a distributed system and can be pushed to its scaling limits, particularly with high write volume.

Can we simplify operations by replacing yet another system with FoundationDB?

What happens to system design when clients have access to a ZooKeeper ensemble that is easily scaled? And offers not just sequential consistency across replicas, but strict serializability?

This talk will examine the building blocks of a proof-of-concept ZooKeeper layer for FDB. First, we'll compare the data models, consistency guarantees, and watch semantics of ZK vs FDB. Then, we'll dig into the specifics of how to emulate ZK by leveraging the Directory Layer for the data model, changefeeds for ZK watches, and evaluate different approaches to essential components like ephemeral nodes and session management.

Speakers
avatar for Paul Hemberger

Paul Hemberger

Tech Lead, HubSpot
Paul Hemberger is a Tech Lead at HubSpot, located in Cambridge, MA. His primary concerns are the correctness, reliability and performance of the CRM, with a particular focus on list segmentation and reporting.



Monday November 18, 2019 2:50pm - 3:10pm PST
Grand Ballroom 8
  Building Layers
  • Session Slides Included Yes

2:50pm PST

NuGraph: GraphDB as a Cloud Service Built Upon JanusGraph and FoundationDB - Jun Li & Hieu Nguyen, eBay
In eBay, we have many applications that require Graph Databases to capture and update business insights from diverse data sources and have insights consumed at real-time. Often such data is sensitive and requires secure protection. Some of these applications can have billions of nodes and billions of edges.  We have built a GraphDB cloud service called NuGraph, which is based on JanusGraph. FoundationDB is chosen as the JanusGraph’s storage plugin, because of its high-performance and distributed transaction support. We will present our GraphDB architecture, and focus on how we deploy and manage FoundationDB in Kubernetes, how we improve JansGraph query performance in a cross-data center environment, how we bulk load the graph into FoundationDB with its transactional support, and how we secure the 3-tier cloud service with limited security support from FoundationDB.

Speakers
avatar for Jun Li

Jun Li

Principal Architect, eBay
Jun Li is currently a Principal Architect at eBay. Since he joined eBay in January 2017, he has been leading the effort on monitoring and self-management of NuData, a geo-distributed transactional document store developed in eBay. The monitoring framework is to provide comprehensive... Read More →
HN

Hieu Nguyen

Applied Researcher, eBay
Hieu Nguyen is an applied researcher at eBay. He is working on developing a consistent, scalable, high-performant and multi-region graph service that is highly demanded by many eBay teams. He graduated from  University of Southern California with a PhD degree in Computer Science... Read More →



Monday November 18, 2019 2:50pm - 3:10pm PST
Grand Ballroom 5
  Real World FDB
  • Session Slides Included Yes

3:10pm PST

Break
Monday November 18, 2019 3:10pm - 3:40pm PST
Grand Ballroom 5 Foyer

3:40pm PST

FoundationDB Record Layer: Open Source Structured Storage on FoundationDB - Nicholas Schiefer, Apple
The FoundationDB Record Layer is an Apache-licensed open source library for managing record-oriented data stores with semantics similar to a relational database on top of the FoundationDB key-value store. It provides a lightweight, highly extensible way to store structured data, including tools for schema management and a rich set of query and indexing features. The Record Layer is a stateless library built for massively multi-tenant cloud environments with a number of features that make it uniquely suited to hosting a large number of databases. As the most mature FoundationDB layer, the Record Layer is a core component of CloudKit, Apple's cloud structured storage system. The Record layer is actively developed by a team at Apple and has a variety of ongoing projects with opportunities for community members to contribute

Speakers
NS

Nicholas Schiefer

Software Engineer, Apple
Nicholas is a developer at Apple working on CloudKit and the FoundationDB Record Layer. His main focus is optimization and execution of declarative queries on structured data.


Monday November 18, 2019 3:40pm - 4:00pm PST
Grand Ballroom 5

4:00pm PST

Build Anything with FoundationDB - Ryan Worl, Ryan Worl
FoundationDB is like a database construction kit, which can be intimidating to new developers and those unfamiliar with database system design and internals. In this talk, Ryan will explain how to start from nothing but an empty key-value store and build anything you can dream up, all while retaining the safety and reliability you expect from FoundationDB.

The first example will focus on the components needed to build a database with a schema, non-blocking schema changes, secondary index construction, and a change data capture system for auditing and ETL.

The second example will show how to add strongly consistent object listing to an eventually consistent object storage system like Amazon S3, as well as adding the ability to append to existing objects, which Amazon S3 doesn't offer.

You'll leave with the ability to map your ideas into keys and values to build creative solutions.

Speakers
avatar for Ryan Worl

Ryan Worl

Consultant, Ryan Worl
Ryan is software consultant primarily focused on improving the performance of online services. His clients use FoundationDB to improve development velocity, provide new features, lower their operational burden, and decrease costs over traditional databases or alternative distributed... Read More →



Monday November 18, 2019 4:00pm - 4:30pm PST
Grand Ballroom 5
  General Session
  • Session Slides Included Yes

4:30pm PST

How We Saved 5x Migrating from Aurora to FoundationDB - Rick Branson & Ray Jenkins, Segment
Segment is on a journey to help companies better understand their customers. A key piece of the Segment solution involves near real-time translation of customer interactions into models of customer Personas. This is the story of how Segment selected FoundationDB to power a critical part of this infrastructure, the identity resolution system.

In this talk, we'll discuss how and why we selected FoundationDB. We'll cover Segment's identity resolution system architecture and how FoundationDB has enabled Segment to deliver Personas modeling at a 5x cost reduction. Finally, we'll discuss the benefits of migrating away from AWS Aurora and the specific operational challenges we faced and how we overcame them.

Speakers
avatar for Ray Jenkins

Ray Jenkins

Software Engineer, Segment
Ray enjoys building fast and fault tolerant distributed systems with a focus on stream processing and near real-time computation. Ray currently works at Segment, where he is also responsible for maintaining Segment's FoundationDB deployment. Prior to Segment, Ray worked at Librato... Read More →
RB

Rick Branson

Software Engineer, Segment
Rick has been deploying, operating, and hacking on distributed systems for over a decade. Currently he works at Segment, where he's spent most of his time leading efforts to improve reliability, scalability, and cost-effectiveness. Before Segment, Rick was the first Infrastructure... Read More →



Monday November 18, 2019 4:30pm - 5:00pm PST
Grand Ballroom 5
  General Session
  • Session Slides Included Yes

5:00pm PST

Networking Reception
Monday November 18, 2019 5:00pm - 5:30pm PST
Grand Ballroom 5