Download presentation
Presentation is loading. Please wait.
Published byBonnie Walsh Modified over 6 years ago
1
The Future of The DBA DevOps, the Cloud Paradigm, & the Microsoft Data Platform Stuart R Ainsworth
2
Agenda 1 2 3 4 1 Describe the typical state of database administration
Define & describe DevOps and the cloud computing paradigm 3 Explore (high-level) the Microsoft Data Platform 4 Discuss the implications for data professionals
3
Assumptions About You 1 2 3 4 2 SQL Server experience
Exposure to database admin & architecture 3 Learning-centered 4 Desire to build modern skills
4
About Me 3 Manage IT for financial services company
Former Data Architect, DBA, developer AtlantaMDF Chapter Leader Infrequent blogger: My career trajectory Experience with large (well, used to be large) databases (multi-terabyte data warehouses, relational db’s) Active in the SQL Server community; speaker, chapter leader Tyr to stay abreast of all the new technologies, but…. About 5 years ago, I moved into management, and my career trajectory has shifted more toward understanding people and processes than being a DBA Focus is on solving business problems regardless of the tools; cant understand depth, but better at breadth.
5
4 You have brains in your head And SQL Skills to boot
You'll soar to great heights On the Data Platform too You're on your own, And you know what you know, And YOU are the one who'll decide where to go. PASS Summit 2016 Lightning Talk that was well received Stepped through the data platform and how skills will shift for DBA’s
6
5 Where Are We? Typical SQL Server Person
7
Typical SQL Server Person
6 Specialized knowledge May have exposure to other infrastructure or development tools Usually reports to Operations (not Development) Often one of the “last stops” in deployment chain You may encounter “the development DBA”; but in most cases, that hat is worn by production dba’s that do some development OR a application developer that writes SQL OR large organizations. This is just supposed to give a representation that’s middle of the road
8
Typical SQL Server Person
7 Development Skills Administration Skills SQL (DDL & DML) Performance Tuning (Code) Index Analysis Data Warehousing Reporting Server Configuration Performance Tuning (Server) Index Maintenance Backups and Restores Security
9
MCSA SQL 2012\2014 8 70-461: Querying Microsoft SQL Server 2012/2014
70-462: Administering Microsoft SQL Server 2012/2014 Databases 70-463: Implementing a Data Warehouse with Microsoft SQL Server 2012/2014
10
What’s coming… 9 Data production is accelerating Data is diversifying
Est .79ZB in 2009 Est 7.9ZB in 2015 Est 35ZB in 2020 (44 times greater than 2009) Data is diversifying Relational Data Big (Size) Data Fast Data Dark Data Lost Data New Data It’s overwhelmening Increase in both size and diversity of data just increases the likelihood that we’ll lose control of the data explosion
11
10 DevOps Philosophy, not Methodology
12
A (Very) Brief Overview
11 DevOps is focused on delivering quality, faster. Philosophical approach, not methodological Automation, infrastructure as code, continuous deployment Emphasis on communication; silo reduction Born out of Agile, several innovators contributing Patrick Debois & Andrew Clay Shafer – Agile Infrastructure (Agile 08) John Allspaw & Paul Hammond – 10+ Deploys Per Day (Velocity 09) Gene Kim, Kevin Behr, & George Spafford – The Phoenix Project Philosophy in that there are a set of general guidelines, but no specific “how to” manual. Embraces several different methods (Scrum, Safe, Kanban), but none are required.
13
The Phoenix Project 12 In simplified terms, this is the way most deployments SHOULD work. However, the more complicated the deployment (the more moving parts, more types), the more likely it is that the deployment will fail, and then you face the dreaded ROLLBACK For example, say you have a .NET web app that connects to a database; when you make a change to the app (system engineers), you also need to deploy some new stored procedures (DBA’s) The simpler the deployment, the less likely the need to rollback.
14
The Phoenix Project 13 AMPLIFY – make louder
Get Feedback into the hands of development as soon as possible, in form of stated issues. AUTOMATE YOUR FEEDBACK SYSTEMS AS MUCH AS POSSIBLE. Data is Data; start looking at feedback as a data challenge to solve. Note that the arrows are circular; if you can speed up the deployment time, and the feedback time, it becomes much easier to move forward and not rollback.
15
The Phoenix Project 14 IF you get your deployment and feedback cycles tight, and eliminate rollbacks, that gives you the opportunity to take risks and expirmenet. Get new features into the hands of beta customers, faster, for example If your application monitoring systems tell you it’s breaking, then you can quickly deploy a fix.
16
Key Takeaways 15 DevOps is rooted in a sense of continuous improvement
People over tools Reduce silos by focusing on shared goals, not technology 1. 2. Lots of great tools: Atlassian products, Slack, HipChat, VictorOps, PagerDuty 3. Tool specialization can create silos; for example, SQL Server DBA’s don’t often hang out with Oracle DBA’s. But end users just want data; they don’t care about the underlying differences. DevOps is about figuring out innovative ways to create value, faster; what we need is a paradigm shift.
17
16 The Cloud Paradigm Infrastructure, Platform, Software
18
What do we mean by “The Cloud”?
17 Trendy marketing term? Network hosting? Internet connected services? Distributed, scalable, shared computing resources?
19
Product-Focused Paradigm
18 Product-focused paradigm; different teams specialize in different products, even when products have similar functionality
20
Cloud Paradigm 19 Applications Data Runtime Middleware O/S
Virtualization Servers Storage Networking Cloud Paradigm looks at function, not products Each level on the stack is dependent on the previous one, but each level is opaque. IE, the OS you chooses shouldn’t care about the underlying Virtual configurations (VMWare vs VirtualBox vs Hyper-V).
21
Cloud Paradigm 20 This is where it gets interesting; you can then begin to group elements in the stack, and consume the management of those groups. On-prem, you manage it all Or you could consume infrastructure as a service from someone else…. Your clients that are running your software don’t manage anything; they just bring their input (e.g., Power BI, Google Docs, etc) - SaaS
22
Cloud Paradigm 21
23
Cloud Paradigm 22
24
Limoncelli, Chalup, Hogan
23 Limoncelli, Chalup, Hogan
25
Ideal System Architecture
24 Scalable Resilient to Failure (Redundant) Service-Oriented Architecture Automated Monitoring, Configuration and Build “Infrastructure As Code”
26
Ideal Release Process 25 Completely Automatic
Code checked in -> new build Unit & regression testing User acceptance testing Continuous Integration Dependent on “infrastructure as code” Micro-releases (100 deployments per day) No rollbacks Micro-releases, and no rollbacks. Sounds familiar, I hope
27
Ideal Operations 26 Automatic instrumentation (Logging)
Long Term Storage Predictive Analytics Automatic Error Logging & Alerting Respond to Every Error On-call Rotation includes Developers Automatic Scaling Scale Up Scale Down “Zero Maintenance” Automatic instrumentation
28
Ideal Data Architecture
28 Consistent SQL Server Relational Engine Hadoop Consistency - all nodes are guaranteed to see same data Availability – every request receives feedback for success/failure Partition Tolerant – system operates despite loss of part of system At any one time, any two attributes are achievable in combination, but not all three at the same time. Available Cassandra Partition Tolerant CAP Principle (Gilbert & Lynch)
29
Microsoft Data Platform
29 Microsoft Data Platform Diversity in Data
30
Microsoft Data Platform
30 Conceptual; think function instead of tools
31
Microsoft Data Platform
31 On Prem tools include Excel for Ad Hoc queries, DataZen for advanced reporting, SQL Server, with SSIS and Analysis Services
32
Microsoft Data Platform
32 Same basic function in Azure, just slightly different tools. While the interfaces and UI may be different, the fucntions are similar to their on-prem counterparts.
33
Impact on Careers 33 Future Prognostications
So what does this all mean? Fun part about guessing about the future is that I don’t have to demonstrate it, I just have to read the tea leaves and get enough things right to look like a genius.
34
Current Trends 35 Companies are recognizing the value of different kinds of data, and an increased need for analytics. Adoption of Big Data technologies is on the rise Data Science jobs are increasing The Internet (or Fog) of Things is coming. Operational methodologies like Agile and DevOps are pushing companies toward the Cloud Paradigm.
35
Current Trends 36 The Cloud Paradigm with its separation of duties is causing companies to realign resources. Infrastructure Teams Platform Teams Software Teams Operational technologies (virtualization, scripting) are allowing organizations to scale out computing resources with fewer human resources.
36
Cloud Paradigm 34 SQL SERVER Applications Data Runtime Middleware O/S
SOFTWARE Applications Data Runtime Middleware O/S Virtualization Servers Storage Networking SQL SERVER PLATFORM INFRA- STRUCTURE Note that SQL Server bridges two different service buckets; platform and software Runtime: the server functions, ie., true administrative tasks like security and backups Data: the database schemas, data, etc. Data is built on the Runtime.
37
My Predictions 37 Increased segregation between Development and Administration roles. Number of development jobs will increase Diversity of data platform. Need for integration. Data mining and analysis skills. Number of administrative jobs will decrease Infrastructure as code, scripting, virtualization Product specific specialists for initial configuration
38
What does this mean for YOU?
38 Choose your path: Development or Administration Developers have opportunities for breadth: Big Data (Hadoop, HDInsight) Data Science (Statistics, R) Visualizations (Reporting, Power BI) Administrators have opportunities for depth: Always On Infrastructure & Platform impacts Scripting & configurations
39
MCSA SQL 2016 39 Querying Data with Transact-SQL
Developing SQL Databases Administering a SQL Database Infrastructure Provisioning SQL Databases Implementing a SQL Data Warehouse Developing SQL Models
40
40
41
Contact Me Stuart R. Ainsworth Twitter: @codegumbo
41 Stuart R. Ainsworth Blog: Slides available:
42
Continuous Improvement for your career
Be Good at Getting Better
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.