Dontcheff

Archive for the ‘Databases’ Category

The DBA profession beyond autonomous: a database without a DBA is like a tree without roots

In Autonomous, Cloud, Databases, DBA on May 30, 2018 at 19:41

“To make a vehicle autonomous, you need to gather massive streams of data from loads of sensors and cameras and process that data on the fly so that the car can ‘see’ what’s around it” Daniel Lyons

Let me add that the data must be stored somewhere, analyzed by some software, monitored and backed up by someone, and so on and so on…

Top 5 Industry Early Adopters Of Autonomous Systems are: (1) Information Technology: Oracle’s Autonomous Data Warehouse Cloud, (2) Automotive, (3) Manufacturing, (4) Retail and (5) Healthcare.

Being an early adopter of ADWC, I must say that it is probably the best product created by Oracle Corporation. For sure part of Top Five.

This month (May 2018), ComputerWeekly published an article quoting Oracle CEO Mark Hurd that the long-term future of database administrators could be at risk if every enterprise adopts the Oracle 18c autonomous database.

“Hurd said it could take almost a year to get on-premise databases patched, whereas patching was instant with the autonomous version. If everyone had the autonomous database, that would change to instantaneous.”

So where does that leave Oracle DBAs around the world? Possibly in the unemployment queue, at least according to Hurd.

“There are hundreds of thousands of DBAs managing Oracle databases. If all of that moved to the autonomous database, the number would change to zero,” Hurd said at an Oracle media event in Redwood Shores, California.

If you are interested in more detail on this subject, I suggest you read the following articles in the order below:

The Robots are coming by James Anthony: “But surely we’ve been here before? Indeed, a quick Google search brings up the following examples of white papers by Oracle with a reference to the database being self-managing all the way back to 2003.”

Oracle Autonomous Database and the Death of the DBA by Tim Hall: “Myself and many others have been talking about this for over a decade. ”

Death of the DBA, Long Live the DBA by Kellyn Pot’Vin-Gorman: “With DBAs that have been around a while, we know the idea that you don’t need a DBA has been around since Oracle 7, the self-healing database.”

No DBA Required? by Tim Hall: “It will be interesting to see what Oracle actually come up with at the end of all this…”

Self-Driving Databases are Coming: What Next for DBAs? by Maria Colgan: “It’s also important for DBAs to remember that the transition to an autonomous environment is not something that will occur overnight.”

Death of the Oracle DBA (again) by Johanthan Stuart: “Twenty years later I run Claremont’s Managed Services practice and the DBA group is our largest delivery team.”

Don’t Fall For The “Autonomous Database” Distraction by Greg McStravick: a totally different point of view on autonomous databases.

Now, “a picture is worth a thousand words”. Here are 5 screenshots from the Autonomous Data Warehouse Cloud documentation:

1. Who will be creating external tables using the DBMS_CLOUD package?

2. Who will run “alter database property set.. ” in order to create credentials for the Oracle Cloud Infrastructure?

3. Who will restore and recover the database in case of any type of failure? Or failures never happen, right?

4. Who will manage run away SQL with cs_resource_manager and run “alter system kill session”?

5. Who will manage the CBO statistics and add hints?

As of today, we have 4 Exadata choices with Autonomous being by far the best. For data warehouse loads for now. As explained by Alan Zeichick, Autonomous Capabilities Will Make Data Warehouses — And DBAs — More Valuable. “No need for a resume writer: DBAs will still have plenty of work to do.”

So still: a database without a DBA is like a tree without roots.

P.S. Check out the book Human + Machine: Reimagining Work in the Age of AI by Paul R. Daugherty and H. James (Jim) Wilson.

Advertisements

Artificial stupidity as a DBA limitation of artificial intelligence

In Data, Database tuning, Databases, DBA on December 6, 2017 at 07:47

“Artificial intelligence is no match for natural stupidity” ― Albert Einstein

What about introducing Artificial Intelligence into the database to an extent it tunes itself into all possible dimensions?

You have probably either seen the question above or have already asked yourself if that was at all possible. On Ask Tom, John from Guildford wrote the following:

As for Artificial Intelligence, well Artificial Stupidity is more likely to be true. Humanity is not privy to the algorithm for intelligence. Anyone who’s had the pleasure of dealing with machine generated code knows that software is no more capable of writing a cohesive system than it is of becoming self-aware.

Provided you’re not trying to be a cheap alternative to an automaton you just need to think. That one function alone differentiates us from computers, so do more of it. The most sublime software on the planet has an IQ of zero, so outdoing it shouldn’t be all that hard.

Stephen Hawking thinks computers may surpass human intelligence and take over the world. Fear artificial stupidity, not artificial intelligence!

Einstein is credited with saying (but it was probably Alexandre Dumas or Elbert Hubbard who deserve the recognition): “The difference between genius and stupidity is that genius has its limits.”

Explore artificial stupidity (AS) and/or read Charles Wheelan’s book Naked Statistics to understand this kind of AI danger. By the way, according to Woody Allen, 94.5% of all statistics are made up!

So what are the limitations of AI? Jay Liebowitz argues that “if intelligence and stupidity naturally exist, and if AI is said to exist, then is there something that might be called “artificial stupidity?” According to him three of these limitations are:

  • Ability to possess and use common sense
  • Development of deep reasoning systems
  • Ability to easily acquire and update knowledge
  • But does artificial intelligence use a database in order to be an artificial intelligence? Few very interesting answers to that question are give by Douglas Green, Jordan Miller and Ramon Morales, here is a summary:

    Although AI could be built without a database, it would probably be more powerful if a database were added. AI and databases are currently not very well integrated. The database is just a standard tool that the AI uses. However, as AI becomes more advanced, it may become more a part of the database itself.

    I don’t believe you can have an effective Artificial Intelligence without a database or memory structure of some kind.

    While it is theoretically possible to have an artificial intelligence without using a database, it makes things a LOT easier if you can store what the AI knows somewhere convenient.

    As Demystifying Artificial Intelligence explains, AI hass been embedded into some of the most fundamental aspects of data management, making those critical data-driven processes more celeritous and manageable.

    Amazon Mechanical Turk is worth looking into and Oracle are also ready for business with AI.

    Matt Johnson, a Navy pilot turned AI researcher, at a conference this simmer by saying that one of the places we are not making a lot of advances is in that teaming, in that interaction (of humans and AI) – Artificial Stupidity: When Artificial Intelligence + Human = Disaster

    Bottom line: if AI uses a database, then the intelligent database should be at least autonomous and have most tasks automated but not relying on artificial stupidity as a DBA limitation of artificial intelligence. Whatever it means… I do not want to curb your enthusiasm but we need to first fill in the skills gap: we need data engineers who understand databases and data warehouses, infrastructure and tools that span data cleaning, ingestion, security, predictions. And in this aspect Cloud is critical and a big differentiator.

    P.S. Is Artificial Intelligence Progress a Bubble? was published 4 days after this blog post.

    Blockchain for DBAs

    In Data, Databases, DBA on October 30, 2017 at 09:25

    “Instead of putting the taxi driver out of a job, blockhchain puts Uber out of a job and lets the taxi driver work with the customer directly.” – Vitalik Buterin

    A blockchain database consists of two kinds of records: transactions and blocks. Blocks contain the lists of the transactions that are hashed and encoded into a hash (Merkle) tree. The linked blocks form a chain as every block holds the hash pointer to the previous block.

    The blockchain can be stored in a flat file or in a database. For example, the Bitcoin core client stores the blockchain metadata using LevelDB (based on Google’s Bigtable database system).

    The diagram above can be used to create the schema in PostgreSQL. “As far as what DBMS you should put it in”, says Ali Razeghi, “that’s up to your use case. If you want to analyze the transactions/wallet IDs to see some patterns or do BI work I would recommend a relational DB. If you want to setup a live ingest with multiple cryptocoins I would recommend something that doesn’t need the transaction log so a MongoDB solution would be good.”

    If you want to setup a MySQL database: here are 8 easy steps.

    But what is the structure of the block, what does it look like?

    The block has 4 fields:

    1. Block Size: The size of the block in bytes
    2. Block Header: Six fields in the block header
    3. Transaction Counter: How many transactions follow
    4. Transactions: The transactions recorded in this block

    The block header has 6 fields:

    1. Version: A version number to track software/protocol upgrades
    2. Previous Block Hash: A reference to the hash of the previous (parent) block in the chain
    3. Merkle Root: A hash of the root of the merkle tree of this block’s transactions
    4. Timestamp: The approximate creation time of this block (seconds from Unix Epoch)
    5. Difficulty Target: The proof-of-work algorithm difficulty target for this block
    6. Nonce: A counter used for the proof-of-work algorithm

    More details, like for example details on block header hash and block height, can be found here.

    But how about blockchain vs. relational database: Which is right for your application? As you can see, because the term “blockchain” is not clearly defined, you could argue that almost any IT project could be described as using a blockchain.

    It is worth reading Guy Harrison’s article Sealing MongoDB documents on the blockchain. Here is a nice quote: “As a database administrator in the early 1990s, I remember the shock I felt when I realized that the contents of the database files were plain text; I’d just assumed they were encrypted and could only be modified by the database engine acting on behalf of a validated user.”

    The Blockchain technology is a very special kind of a distributed database. Sebastien Meunier’s post concludes that ironically, there is no consensus on the definition of what blockchain technology is.

    I particularly, like his last question: Is a private blockchain without token really more efficient than a centralized system? And I would add: private blockchain, really?

    But once more, what is blockchain? Rockford Lhotka gives a very good DBA-friendly definition/characteristics of blockchain:

    1. A linked list where each node contains data
    2. Immutable:
    – Each new node is cryptographically linked to the previous node
    – The list and the data in each node is therefore immutable, tampering breaks the cryptography
    3. Append-only
    – New nodes can be added to the list, though existing nodes can’t be altered
    4. Persistent
    – Hence it is a data store – the list and nodes of data are persisted
    5. Distributed
    – Copies of the list exist on many physical devices/servers
    – Failure of 1+ physical devices has no impact on the integrity of the data
    – The physical devices form a type of networked cluster and work together
    – New nodes are only appended to the list if some quorum of physical devices agree with the cryptography and validity of the node via consistent algorithms running on all devices.

    Kevin Ford’s reply is a good one to conclude with: “Based on this description (above) it really sounds like your (Rockford Lhotka’s) earlier comparison to the hype around XML is spot on. It sounds like in and of itself it isn’t particularly anything except a low level technology until you structure it to meet a particular problem.”

    The nature of blockchain technology makes it difficult to work with high transnational volumes.

    But DBAs can have a look at (1) BigchainDB, a database with several blockchain characteristics added: high-transaction, decentralized database, immutability & native support for assets and (2) at Chainfrog if interested in connecting legacy databases together. As far as I know, they support as of now at least MySQL and SQL Server.